0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:AnalyzeTextFilesData.pl</title>
|
|
4 <meta http-equiv="content-type" content="text/html;charset=utf-8">
|
|
5 <link rel="stylesheet" type="text/css" href="../../../css/MayaChemToolsCode.css">
|
|
6 </head>
|
|
7 <body leftmargin="20" rightmargin="20" topmargin="10" bottommargin="10">
|
|
8 <br/>
|
|
9 <center>
|
|
10 <a href="http://www.mayachemtools.org" title="MayaChemTools Home"><img src="../../../images/MayaChemToolsLogo.gif" border="0" alt="MayaChemTools"></a>
|
|
11 </center>
|
|
12 <br/>
|
|
13 <pre>
|
|
14 1 #!/usr/bin/perl -w
|
|
15 2 <span class="c">#</span>
|
|
16 3 <span class="c"># $RCSfile: AnalyzeTextFilesData.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:04 $</span>
|
|
18 5 <span class="c"># $Revision: 1.36 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$FindBin::Bin/../lib"</span><span class="sc">;</span>
|
|
44 31 <span class="k">use</span> <span class="w">Getopt::Long</span><span class="sc">;</span>
|
|
45 32 <span class="k">use</span> <span class="w">File::Basename</span><span class="sc">;</span>
|
|
46 33 <span class="k">use</span> <span class="w">Text::ParseWords</span><span class="sc">;</span>
|
|
47 34 <span class="k">use</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
48 35 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
49 36 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
50 37 <span class="k">use</span> <span class="w">StatisticsUtil</span><span class="sc">;</span>
|
|
51 38
|
|
52 39 <span class="k">my</span><span class="s">(</span><span class="i">$ScriptName</span><span class="cm">,</span> <span class="i">%Options</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="cm">,</span> <span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$TotalTime</span><span class="s">)</span><span class="sc">;</span>
|
|
53 40
|
|
54 41 <span class="c"># Autoflush STDOUT</span>
|
|
55 42 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
56 43
|
|
57 44 <span class="c"># Starting message...</span>
|
|
58 45 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span>
|
|
59 46 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span>
|
|
60 47 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
61 48
|
|
62 49 <span class="c"># Get the options and setup script...</span>
|
|
63 50 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
64 51 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
65 52 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">"$FindBin::Bin/$ScriptName"</span><span class="s">)</span><span class="sc">;</span>
|
|
66 53 <span class="s">}</span>
|
|
67 54
|
|
68 55 <span class="k">my</span><span class="s">(</span><span class="i">@TextFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
69 56 <span class="i">@TextFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">"csv tsv"</span><span class="s">)</span><span class="sc">;</span>
|
|
70 57
|
|
71 58 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
72 59 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
73 60 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
74 61
|
|
75 62 <span class="c"># Collect column information for all the text files...</span>
|
|
76 63 <span class="k">print</span> <span class="q">"Checking input text file(s)...\n"</span><span class="sc">;</span>
|
|
77 64 <span class="k">my</span><span class="s">(</span><span class="i">%TextFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
78 65 <span class="i">RetrieveTextFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
79 66 <span class="i">ProcessColumnsInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
80 67
|
|
81 68 <span class="c"># Generate output files...</span>
|
|
82 69 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
83 70 <span class="k">if</span> <span class="s">(</span><span class="i">@TextFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
84 71 <span class="k">print</span> <span class="q">"\nProcessing text files...\n"</span><span class="sc">;</span>
|
|
85 72 <span class="s">}</span>
|
|
86 73 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
87 74 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span>
|
|
88 75 <span class="k">print</span> <span class="q">"\nProcessing file $TextFilesList[$FileIndex]...\n"</span><span class="sc">;</span>
|
|
89 76 <span class="i">AnalyzeTextFile</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
90 77 <span class="s">}</span>
|
|
91 78 <span class="s">}</span>
|
|
92 79 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span>
|
|
93 80
|
|
94 81 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
95 82 <span class="i">$TotalTime</span> = <span class="w">timediff</span> <span class="s">(</span><span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="s">)</span><span class="sc">;</span>
|
|
96 83 <span class="k">print</span> <span class="q">"Total time: "</span><span class="cm">,</span> <span class="i">timestr</span><span class="s">(</span><span class="i">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
97 84
|
|
98 85 <span class="c">###############################################################################</span>
|
|
99 86
|
|
100 87 <span class="c"># Analyze data...</span>
|
|
101 <a name="AnalyzeTextFile-"></a> 88 <span class="k">sub </span><span class="m">AnalyzeTextFile</span> <span class="s">{</span>
|
|
102 89 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
103 90 <span class="k">my</span><span class="s">(</span><span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
104 91
|
|
105 92 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
106 93 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
107 94 <span class="i">@ColNumsToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">UniqueColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
108 95 <span class="i">%ColValuesToAnalyzeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
109 96 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
110 97 <span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMap</span>{<span class="i">$ColNum</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
111 98 <span class="s">}</span>
|
|
112 99
|
|
113 100 <span class="k">my</span><span class="s">(</span><span class="i">$LineCount</span><span class="cm">,</span> <span class="i">$InvalidLineCount</span><span class="cm">,</span> <span class="i">@InvalidColLabels</span><span class="s">)</span><span class="sc">;</span>
|
|
114 101
|
|
115 102 <span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $TextFile: $! \n"</span><span class="sc">;</span>
|
|
116 103 <span class="c"># Skip over column labels line in text file and collect appropriate column data</span>
|
|
117 104 <span class="c"># for analysis...</span>
|
|
118 105 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
119 106 <span class="i">$LineCount</span> = <span class="n">1</span><span class="sc">;</span>
|
|
120 107 <span class="i">$InvalidLineCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
121 108 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
122 109 <span class="i">$LineCount</span>++<span class="sc">;</span>
|
|
123 110 <span class="i">@LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
124 111 <span class="i">@InvalidColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
125 112 <span class="j">COLNUM:</span> <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
126 113 <span class="i">$Value</span> = <span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
127 114 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>}<span class="s">)</span> <span class="s">{</span>
|
|
128 115 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsNumerical</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
129 116 <span class="k">push</span> <span class="i">@InvalidColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
130 117 <span class="k">next</span> <span class="j">COLNUM</span><span class="sc">;</span>
|
|
131 118 <span class="s">}</span>
|
|
132 119 <span class="s">}</span>
|
|
133 120 <span class="k">push</span> <span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMap</span>{<span class="i">$ColNum</span>}}<span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
134 121 <span class="s">}</span>
|
|
135 122 <span class="k">if</span> <span class="s">(</span><span class="i">@InvalidColLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
136 123 <span class="i">$InvalidLineCount</span>++<span class="sc">;</span>
|
|
137 124 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >=<span class="n">4</span> <span class="s">)</span> <span class="s">{</span>
|
|
138 125 <span class="k">print</span> <span class="q">"Line number $LineCount contains "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@InvalidColLabels</span><span class="s">)</span><span class="cm">,</span><span class="q">" non-numerical or empty value(s) for column(s) - "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@InvalidColLabels</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span><span class="q">" - to be analyzed: $Line \n"</span><span class="sc">;</span>
|
|
139 126 <span class="s">}</span>
|
|
140 127 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">3</span><span class="s">)</span> <span class="s">{</span>
|
|
141 128 <span class="k">print</span> <span class="q">"Line number $LineCount contains "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@InvalidColLabels</span><span class="s">)</span><span class="cm">,</span><span class="q">" non-numerical or empty value(s) for column(s) - "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@InvalidColLabels</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span><span class="q">" - to be analyzed...\n"</span><span class="sc">;</span>
|
|
142 129 <span class="s">}</span>
|
|
143 130 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
144 131 <span class="k">print</span> <span class="q">"Line number $LineCount contains "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@InvalidColLabels</span><span class="s">)</span><span class="cm">,</span><span class="q">" non-numerical or empty value(s) for columns to be analyzed...\n"</span><span class="sc">;</span>
|
|
145 132 <span class="s">}</span>
|
|
146 133 <span class="s">}</span>
|
|
147 134 <span class="s">}</span>
|
|
148 135 <span class="k">if</span> <span class="s">(</span><span class="i">$InvalidLineCount</span> && <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
149 136 <span class="k">print</span> <span class="q">"Non-numerical or empty data present in $InvalidLineCount line(s)...\n"</span><span class="sc">;</span>
|
|
150 137 <span class="s">}</span>
|
|
151 138 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span>
|
|
152 139
|
|
153 140 <span class="c"># Perform the analysis...</span>
|
|
154 141 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedFunctionNames</span><span class="cm">,</span> <span class="i">$SpecifiedFunction</span><span class="s">)</span><span class="sc">;</span>
|
|
155 142 <span class="i">@SpecifiedFunctionNames</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
156 143
|
|
157 144 <span class="k">for</span> <span class="i">$SpecifiedFunction</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
158 145 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> !~ <span class="q">/^(Covariance|Correlation|Frequency|Rsquare|StandardScores|StandardScoresN)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
159 146 <span class="k">push</span> <span class="i">@SpecifiedFunctionNames</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span>}<span class="sc">;</span>
|
|
160 147 <span class="s">}</span>
|
|
161 148 <span class="s">}</span>
|
|
162 149 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedFunctionNames</span><span class="s">)</span> <span class="s">{</span>
|
|
163 150 <span class="i">PerformAnalysis</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> \<span class="i">@SpecifiedFunctionNames</span><span class="cm">,</span> \<span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span>
|
|
164 151 <span class="s">}</span>
|
|
165 152 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>}<span class="s">)</span> || <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>}<span class="s">)</span> || <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
166 153 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AllColumnPairs</span>}<span class="s">)</span> <span class="s">{</span>
|
|
167 154 <span class="i">PerformMatrixAnalysis</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> \<span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
168 155 <span class="s">}</span>
|
|
169 156 <span class="k">else</span> <span class="s">{</span>
|
|
170 157 <span class="c"># Perform pairwise analysis for specified columns and write out calculated values - correlation</span>
|
|
171 158 <span class="c"># rsquare, or covariance - in the same file.</span>
|
|
172 159 <span class="i">PerformColumnPairAnalysis</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> \<span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
173 160 <span class="s">}</span>
|
|
174 161 <span class="s">}</span>
|
|
175 162 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">standardscores</span>}<span class="s">)</span> || <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">standardscoresn</span>}<span class="s">)</span> <span class="s">)</span> <span class="s">{</span>
|
|
176 163 <span class="i">PerformStandardScoresAnalysis</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> \<span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
177 164 <span class="s">}</span>
|
|
178 165 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">frequency</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
179 166 <span class="i">PerformFrequencyAnalysis</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> \<span class="i">%ColValuesToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
180 167 <span class="s">}</span>
|
|
181 168 <span class="s">}</span>
|
|
182 169
|
|
183 170 <span class="c"># Calculate values for various statistical functions...</span>
|
|
184 <a name="PerformAnalysis-"></a> 171 <span class="k">sub </span><span class="m">PerformAnalysis</span> <span class="s">{</span>
|
|
185 172 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$SpecifiedFunctionNamesRef</span><span class="cm">,</span> <span class="i">$ColValuesToAnalyzeMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
186 173 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$SpecifiedFunction</span><span class="cm">,</span> <span class="i">$Label</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="s">)</span><span class="sc">;</span>
|
|
187 174
|
|
188 175 <span class="i">$NewTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} . <span class="q">"."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
189 176
|
|
190 177 <span class="k">print</span> <span class="q">"Generating new text file $NewTextFile...\n"</span><span class="sc">;</span>
|
|
191 178 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$NewTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewTextFile: $! \n"</span><span class="sc">;</span>
|
|
192 179
|
|
193 180 <span class="c"># Write out column labels...</span>
|
|
194 181 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
195 182 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"ColumnID"</span><span class="sc">;</span>
|
|
196 183 <span class="k">for</span> <span class="i">$SpecifiedFunction</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SpecifiedFunctionNamesRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
197 184 <span class="i">$Label</span> = <span class="i">$SpecifiedFunction</span><span class="sc">;</span>
|
|
198 185 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^(KLargest|KSmallest)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
199 186 <span class="k">my</span><span class="s">(</span><span class="i">$KthValue</span><span class="s">)</span><span class="sc">;</span>
|
|
200 187 <span class="i">$KthValue</span> = <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^KLargest$/i</span><span class="s">)</span> ? <span class="i">$OptionsInfo</span>{<span class="w">KLargest</span>} <span class="co">:</span> <span class="i">$OptionsInfo</span>{<span class="w">KSmallest</span>}<span class="sc">;</span>
|
|
201 188 <span class="i">$Label</span> = <span class="i">AddNumberSuffix</span><span class="s">(</span><span class="i">$KthValue</span><span class="s">)</span> . <span class="q">"$SpecifiedFunction"</span><span class="sc">;</span>
|
|
202 189 <span class="i">$Label</span> =~ <span class="q">s/K//g</span><span class="sc">;</span>
|
|
203 190 <span class="s">}</span>
|
|
204 191 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^TrimMean$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
205 192 <span class="i">$Label</span> = <span class="q">"${SpecifiedFunction}($OptionsInfo{TrimFraction})"</span><span class="sc">;</span>
|
|
206 193 <span class="s">}</span>
|
|
207 194 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$Label</span><span class="sc">;</span>
|
|
208 195 <span class="s">}</span>
|
|
209 196 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
210 197 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
211 198
|
|
212 199 <span class="c"># Go over each column to be analyzed...</span>
|
|
213 200 <span class="i">@ColNumsToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
214 201
|
|
215 202 <span class="c"># Turn off "strict"; otherwise, invoking statistical functions using function name string</span>
|
|
216 203 <span class="c"># is problematic.</span>
|
|
217 204 <span class="k">no</span> <span class="w">strict</span><span class="sc">;</span>
|
|
218 205
|
|
219 206 <span class="k">my</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">%CalculatedValues</span><span class="s">)</span><span class="sc">;</span>
|
|
220 207 <span class="i">%CalculatedValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
221 208 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
222 209 <span class="i">@RowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
223 210 <span class="c"># Setup column id...</span>
|
|
224 211 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
225 212 <span class="i">$ColValuesRef</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="sc">;</span>
|
|
226 213 <span class="j">FUNCTIONNAME:</span> <span class="k">for</span> <span class="i">$SpecifiedFunction</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$SpecifiedFunctionNamesRef</span>}<span class="s">)</span> <span class="s">{</span>
|
|
227 214 <span class="i">$Value</span> = <span class="q">""</span><span class="sc">;</span>
|
|
228 215 <span class="k">if</span> <span class="s">(</span>!<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
229 216 <span class="c"># Invalid column values...</span>
|
|
230 217 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
231 218 <span class="k">next</span> <span class="j">FUNCTIONNAME</span><span class="sc">;</span>
|
|
232 219 <span class="s">}</span>
|
|
233 220 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^Count$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
234 221 <span class="i">$Value</span> = <span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="sc">;</span>
|
|
235 222 <span class="s">}</span>
|
|
236 223 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^KLargest$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
237 224 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">KLargest</span>}<span class="s">)</span><span class="sc">;</span>
|
|
238 225 <span class="s">}</span>
|
|
239 226 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^KSmallest$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
240 227 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">KSmallest</span>}<span class="s">)</span><span class="sc">;</span>
|
|
241 228 <span class="s">}</span>
|
|
242 229 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^StandardDeviation$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
243 230 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
244 231 <span class="i">$Value</span> = <span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>}<span class="sc">;</span>
|
|
245 232 <span class="s">}</span>
|
|
246 233 <span class="k">else</span> <span class="s">{</span>
|
|
247 234 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
248 235 <span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
249 236 <span class="s">}</span>
|
|
250 237 <span class="s">}</span>
|
|
251 238 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^StandardError$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
252 239 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
253 240 <span class="i">$Value</span> = <span class="i">StandardDeviation</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
254 241 <span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>} = <span class="i">$Value</span><span class="sc">;</span>
|
|
255 242 <span class="s">}</span>
|
|
256 243 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>}<span class="s">)</span> <span class="s">{</span>
|
|
257 244 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$CalculatedValues</span>{<span class="i">$ColNum</span>}{<span class="w">StandardDeviation</span>}<span class="cm">,</span> <span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="s">)</span><span class="sc">;</span>
|
|
258 245 <span class="s">}</span>
|
|
259 246 <span class="s">}</span>
|
|
260 247 <span class="k">elsif</span> <span class="s">(</span><span class="i">$SpecifiedFunction</span> =~ <span class="q">/^TrimMean$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
261 248 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">TrimFraction</span>}<span class="s">)</span><span class="sc">;</span>
|
|
262 249 <span class="s">}</span>
|
|
263 250 <span class="k">else</span> <span class="s">{</span>
|
|
264 251 <span class="i">$Value</span> = <span class="i">&$SpecifiedFunction</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
265 252 <span class="s">}</span>
|
|
266 253 <span class="c"># Format the output value. And add zero to get rid of tariling zeros...</span>
|
|
267 254 <span class="i">$Value</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
268 255 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
269 256 <span class="s">}</span>
|
|
270 257 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
271 258 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
272 259 <span class="s">}</span>
|
|
273 260 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
274 261 <span class="s">}</span>
|
|
275 262
|
|
276 263 <span class="c"># Calculate covariance, correlation, rsquare for specified column pairs....</span>
|
|
277 <a name="PerformColumnPairAnalysis-"></a> 264 <span class="k">sub </span><span class="m">PerformColumnPairAnalysis</span> <span class="s">{</span>
|
|
278 265 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$ColValuesToAnalyzeMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
279 266 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$CalculateCorrelation</span><span class="cm">,</span> <span class="i">$CalculateRSquare</span><span class="cm">,</span> <span class="i">$CalculateCovariance</span><span class="s">)</span><span class="sc">;</span>
|
|
280 267 <span class="i">$CalculateCorrelation</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
281 268 <span class="i">$CalculateRSquare</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
282 269 <span class="i">$CalculateCovariance</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
283 270
|
|
284 271 <span class="i">$NewTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="q">"ColumnPairsAnalysis."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
285 272 <span class="k">print</span> <span class="q">"Generating new text file $NewTextFile...\n"</span><span class="sc">;</span>
|
|
286 273 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$NewTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewTextFile: $! \n"</span><span class="sc">;</span>
|
|
287 274
|
|
288 275 <span class="c"># Write out the column labels...</span>
|
|
289 276 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
290 277 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="s">(</span><span class="q">"ColumnID1"</span><span class="cm">,</span> <span class="q">"ColumnID2"</span><span class="s">)</span><span class="sc">;</span>
|
|
291 278 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
292 279 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"Correlation"</span><span class="sc">;</span>
|
|
293 280 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
294 281 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"RSquare"</span><span class="sc">;</span>
|
|
295 282 <span class="s">}</span>
|
|
296 283 <span class="s">}</span>
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
298 285 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"Covariance"</span><span class="sc">;</span>
|
|
299 286 <span class="s">}</span>
|
|
300 287 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
301 288 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
302 289
|
|
303 290 <span class="c"># Go over each column pair...</span>
|
|
304 291 <span class="k">my</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="cm">,</span> <span class="i">$RSquareValue</span><span class="cm">,</span> <span class="i">$CovarianceValue</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="cm">,</span> <span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="cm">,</span> <span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="cm">,</span> <span class="i">@ColPairs1ToAnalyze</span><span class="cm">,</span> <span class="i">@ColPairs2ToAnalyze</span><span class="cm">,</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="s">)</span><span class="sc">;</span>
|
|
305 292
|
|
306 293 <span class="i">@ColPairs1ToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs1ToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
307 294 <span class="i">@ColPairs2ToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs2ToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
308 295 <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#ColPairs1ToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
309 296 <span class="i">@RowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
310 297 <span class="i">$ColNum1</span> = <span class="i">$ColPairs1ToAnalyze</span>[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
311 298 <span class="i">$ColNum2</span> = <span class="i">$ColPairs2ToAnalyze</span>[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
312 299 <span class="i">$ColValuesRef1</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum1</span>}}<span class="sc">;</span>
|
|
313 300 <span class="i">$ColValuesRef2</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum2</span>}}<span class="sc">;</span>
|
|
314 301
|
|
315 302 <span class="c"># Setup column ids...</span>
|
|
316 303 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum1</span>]<span class="sc">;</span>
|
|
317 304 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum2</span>]<span class="sc">;</span>
|
|
318 305
|
|
319 306 <span class="k">if</span> <span class="s">(</span><span class="i">@$ColValuesRef1</span> != <span class="i">@$ColValuesRef2</span><span class="s">)</span> <span class="s">{</span>
|
|
320 307 <span class="c"># Print a warning...</span>
|
|
321 308 <span class="k">warn</span> <span class="q">"Warning: Skipping analysis for column pair $TextFilesInfo{ColLabels}[$Index][$ColNum1], $TextFilesInfo{ColLabels}[$Index][$ColNum2]: Number of valid data values must be same.\n"</span><span class="sc">;</span>
|
|
322 309 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
323 310 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="q">""</span><span class="sc">;</span>
|
|
324 311 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
325 312 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="q">""</span><span class="sc">;</span>
|
|
326 313 <span class="s">}</span>
|
|
327 314 <span class="s">}</span>
|
|
328 315 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
329 316 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="q">""</span><span class="sc">;</span>
|
|
330 317 <span class="s">}</span>
|
|
331 318 <span class="s">}</span>
|
|
332 319 <span class="k">else</span> <span class="s">{</span>
|
|
333 320 <span class="c"># Calculate appropriate value...</span>
|
|
334 321 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
335 322 <span class="i">$CorrelationValue</span> = <span class="i">Correlation</span><span class="s">(</span><span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="s">)</span><span class="sc">;</span>
|
|
336 323 <span class="i">$Value</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$CorrelationValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
337 324 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
338 325 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
339 326 <span class="i">$RSquareValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="i">$CorrelationValue</span> ** <span class="n">2</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
340 327 <span class="i">$Value</span> = <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$RSquareValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$RSquareValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
341 328 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
342 329 <span class="s">}</span>
|
|
343 330 <span class="s">}</span>
|
|
344 331 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
345 332 <span class="i">$CovarianceValue</span> = <span class="i">Covariance</span><span class="s">(</span><span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="s">)</span><span class="sc">;</span>
|
|
346 333 <span class="i">$Value</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CovarianceValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CovarianceValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$CovarianceValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
347 334 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
348 335 <span class="s">}</span>
|
|
349 336 <span class="s">}</span>
|
|
350 337 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
351 338 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
352 339 <span class="s">}</span>
|
|
353 340 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
354 341 <span class="s">}</span>
|
|
355 342
|
|
356 343 <span class="c"># Generate histogram numbers...</span>
|
|
357 <a name="PerformFrequencyAnalysis-"></a> 344 <span class="k">sub </span><span class="m">PerformFrequencyAnalysis</span> <span class="s">{</span>
|
|
358 345 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$ColValuesToAnalyzeMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
359 346 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$BinValue</span><span class="cm">,</span> <span class="i">$FrequencyValue</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">%FrequencyMap</span><span class="s">)</span><span class="sc">;</span>
|
|
360 347
|
|
361 348 <span class="i">@ColNumsToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
362 349 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
363 350 <span class="i">$NewTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>] . <span class="q">"FrequencyAnalysis."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
364 351 <span class="k">print</span> <span class="q">"Generating new text file $NewTextFile...\n"</span><span class="sc">;</span>
|
|
365 352 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$NewTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewTextFile: $! \n"</span><span class="sc">;</span>
|
|
366 353
|
|
367 354 <span class="c"># Write out the column labels...</span>
|
|
368 355 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
369 356 <span class="k">push</span> <span class="i">@ColLabels</span> <span class="cm">,</span> <span class="s">(</span><span class="q">"Bins"</span><span class="cm">,</span> <span class="q">"Frequency"</span><span class="s">)</span><span class="sc">;</span>
|
|
370 357 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
371 358 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
372 359
|
|
373 360 <span class="c">#Calculate and write out frequency values...</span>
|
|
374 361 <span class="i">%FrequencyMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
375 362 <span class="i">$ColValuesRef</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="sc">;</span>
|
|
376 363 <span class="k">if</span> <span class="s">(</span><span class="i">@$ColValuesRef</span><span class="s">)</span> <span class="s">{</span>
|
|
377 364 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
378 365 <span class="i">%FrequencyMap</span> = <span class="i">Frequency</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> \<span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="s">)</span><span class="sc">;</span>
|
|
379 366 <span class="s">}</span>
|
|
380 367 <span class="k">else</span> <span class="s">{</span>
|
|
381 368 <span class="i">%FrequencyMap</span> = <span class="i">Frequency</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>}<span class="s">)</span><span class="sc">;</span>
|
|
382 369 <span class="s">}</span>
|
|
383 370 <span class="s">}</span>
|
|
384 371 <span class="k">for</span> <span class="i">$BinValue</span> <span class="s">(</span><span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="k">keys</span> <span class="i">%FrequencyMap</span><span class="s">)</span> <span class="s">{</span>
|
|
385 372 <span class="i">$FrequencyValue</span> = <span class="i">$FrequencyMap</span>{<span class="i">$BinValue</span>}<span class="sc">;</span>
|
|
386 373
|
|
387 374 <span class="i">@RowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
388 375 <span class="i">$Value</span> = <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$BinValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$BinValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
389 376 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
390 377 <span class="i">$Value</span> = <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$FrequencyValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$FrequencyValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
391 378 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
392 379
|
|
393 380 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
394 381 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
395 382 <span class="s">}</span>
|
|
396 383 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
397 384 <span class="s">}</span>
|
|
398 385 <span class="s">}</span>
|
|
399 386
|
|
400 387 <span class="c"># Calculate covariance, correlation/rsquare matrices....</span>
|
|
401 <a name="PerformMatrixAnalysis-"></a> 388 <span class="k">sub </span><span class="m">PerformMatrixAnalysis</span> <span class="s">{</span>
|
|
402 389 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$ColValuesToAnalyzeMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
403 390 <span class="k">my</span><span class="s">(</span><span class="i">$CorrelationTextFile</span><span class="cm">,</span> <span class="i">$CovarianceTextFile</span><span class="cm">,</span> <span class="i">$RSquareTextFile</span><span class="cm">,</span> <span class="i">$CalculateCorrelation</span><span class="cm">,</span> <span class="i">$CalculateRSquare</span><span class="cm">,</span> <span class="i">$CalculateCovariance</span><span class="s">)</span><span class="sc">;</span>
|
|
404 391
|
|
405 392 <span class="i">$CalculateCorrelation</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
406 393 <span class="i">$CalculateRSquare</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
407 394 <span class="i">$CalculateCovariance</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
408 395
|
|
409 396 <span class="i">$CorrelationTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="q">"CorrelationMatrix."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
410 397 <span class="i">$RSquareTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="q">"RSquareMatrix."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
411 398 <span class="i">$CovarianceTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="q">"CovarianceMatrix."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
412 399
|
|
413 400 <span class="k">my</span><span class="s">(</span><span class="i">$TextFilesList</span><span class="cm">,</span> <span class="i">$Delimiter</span><span class="s">)</span><span class="sc">;</span>
|
|
414 401 <span class="i">$TextFilesList</span> = <span class="q">""</span><span class="sc">;</span>
|
|
415 402 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
416 403 <span class="i">$TextFilesList</span> = <span class="i">$CorrelationTextFile</span><span class="sc">;</span>
|
|
417 404 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
418 405 <span class="i">$TextFilesList</span> .= <span class="q">", $CorrelationTextFile"</span><span class="sc">;</span>
|
|
419 406 <span class="s">}</span>
|
|
420 407 <span class="s">}</span>
|
|
421 408 <span class="i">$Delimiter</span> = <span class="k">length</span><span class="s">(</span><span class="i">$TextFilesList</span><span class="s">)</span> ? <span class="q">","</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
422 409 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
423 410 <span class="i">$TextFilesList</span> .= <span class="q">"${Delimiter} ${CorrelationTextFile}"</span><span class="sc">;</span>
|
|
424 411 <span class="s">}</span>
|
|
425 412 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesList</span> =~ <span class="q">/\,/</span><span class="s">)</span> <span class="s">{</span>
|
|
426 413 <span class="k">print</span> <span class="q">"Generating new text files $TextFilesList...\n"</span>
|
|
427 414 <span class="s">}</span>
|
|
428 415 <span class="k">else</span> <span class="s">{</span>
|
|
429 416 <span class="k">print</span> <span class="q">"Generating new text file $TextFilesList...\n"</span>
|
|
430 417 <span class="s">}</span>
|
|
431 418 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
432 419 <span class="k">open</span> <span class="w">CORRELATIONTEXTFILE</span><span class="cm">,</span> <span class="q">">$CorrelationTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $CorrelationTextFile: $! \n"</span><span class="sc">;</span>
|
|
433 420 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
434 421 <span class="k">open</span> <span class="w">RSQUARETEXTFILE</span><span class="cm">,</span> <span class="q">">$RSquareTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $RSquareTextFile: $! \n"</span><span class="sc">;</span>
|
|
435 422 <span class="s">}</span>
|
|
436 423 <span class="s">}</span>
|
|
437 424 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
438 425 <span class="k">open</span> <span class="w">COVARIANCETEXTFILE</span><span class="cm">,</span> <span class="q">">$CovarianceTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $CovarianceTextFile: $! \n"</span><span class="sc">;</span>
|
|
439 426 <span class="s">}</span>
|
|
440 427
|
|
441 428 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$CorrelationValue</span><span class="cm">,</span> <span class="i">$RSquareValue</span><span class="cm">,</span> <span class="i">$CovarianceValue</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="cm">,</span> <span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@CovarianceRowValues</span><span class="cm">,</span> <span class="i">@CorrelationRowValues</span><span class="cm">,</span> <span class="i">@RSquareRowValues</span><span class="s">)</span><span class="sc">;</span>
|
|
442 429
|
|
443 430 <span class="c"># Write out the column labels...</span>
|
|
444 431 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
445 432 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">""</span><span class="sc">;</span>
|
|
446 433 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
447 434 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
448 435 <span class="s">}</span>
|
|
449 436 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
450 437 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
451 438 <span class="k">print</span> <span class="i">CORRELATIONTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
452 439 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
453 440 <span class="k">print</span> <span class="i">RSQUARETEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
454 441 <span class="s">}</span>
|
|
455 442 <span class="s">}</span>
|
|
456 443 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
457 444 <span class="k">print</span> <span class="i">COVARIANCETEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
458 445 <span class="s">}</span>
|
|
459 446
|
|
460 447 <span class="c"># Due to symmetric nature of these matrices, only one half needs to be</span>
|
|
461 448 <span class="c"># calculated. So, just calculate the lower half and copy it to upper half...</span>
|
|
462 449 <span class="k">my</span><span class="s">(</span><span class="i">%CorrelationMatrixMap</span><span class="cm">,</span> <span class="i">%RSquareMatrixMap</span><span class="cm">,</span> <span class="i">%CovarianceMatrixMap</span><span class="s">)</span><span class="sc">;</span>
|
|
463 450
|
|
464 451 <span class="i">%CorrelationMatrixMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">%RSquareMatrixMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> <span class="i">%CovarianceMatrixMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
465 452 <span class="k">for</span> <span class="i">$ColNum1</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
466 453 <span class="k">for</span> <span class="i">$ColNum2</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$ColNum1</span><span class="s">)</span> <span class="s">{</span>
|
|
467 454 <span class="i">$ColValuesRef1</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum1</span>}}<span class="sc">;</span>
|
|
468 455 <span class="i">$ColValuesRef2</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum2</span>}}<span class="sc">;</span>
|
|
469 456 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
470 457 <span class="i">$CorrelationValue</span> = <span class="i">Correlation</span><span class="s">(</span><span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="s">)</span><span class="sc">;</span>
|
|
471 458 <span class="i">$CorrelationValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$CorrelationValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
472 459 <span class="i">$CorrelationMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>} = <span class="i">$CorrelationValue</span><span class="sc">;</span>
|
|
473 460 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum1</span> != <span class="i">$ColNum2</span><span class="s">)</span> <span class="s">{</span>
|
|
474 461 <span class="i">$CorrelationMatrixMap</span>{<span class="i">$ColNum2</span>}{<span class="i">$ColNum1</span>} = <span class="i">$CorrelationValue</span><span class="sc">;</span>
|
|
475 462 <span class="s">}</span>
|
|
476 463 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
477 464 <span class="i">$RSquareValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CorrelationValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="i">$CorrelationValue</span> ** <span class="n">2</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
478 465 <span class="i">$RSquareValue</span> = <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$RSquareValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$RSquareValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
479 466 <span class="i">$RSquareMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>} = <span class="i">$RSquareValue</span><span class="sc">;</span>
|
|
480 467 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum1</span> != <span class="i">$ColNum2</span><span class="s">)</span> <span class="s">{</span>
|
|
481 468 <span class="i">$RSquareMatrixMap</span>{<span class="i">$ColNum2</span>}{<span class="i">$ColNum1</span>} = <span class="i">$RSquareValue</span><span class="sc">;</span>
|
|
482 469 <span class="s">}</span>
|
|
483 470 <span class="s">}</span>
|
|
484 471 <span class="s">}</span>
|
|
485 472 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
486 473 <span class="i">$CovarianceValue</span> = <span class="i">Covariance</span><span class="s">(</span><span class="i">$ColValuesRef1</span><span class="cm">,</span> <span class="i">$ColValuesRef2</span><span class="s">)</span><span class="sc">;</span>
|
|
487 474 <span class="i">$CovarianceValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$CovarianceValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$CovarianceValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$CovarianceValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
488 475 <span class="i">$CovarianceMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>} = <span class="i">$CovarianceValue</span><span class="sc">;</span>
|
|
489 476 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum1</span> != <span class="i">$ColNum2</span><span class="s">)</span> <span class="s">{</span>
|
|
490 477 <span class="i">$CovarianceMatrixMap</span>{<span class="i">$ColNum2</span>}{<span class="i">$ColNum1</span>} = <span class="i">$CovarianceValue</span><span class="sc">;</span>
|
|
491 478 <span class="s">}</span>
|
|
492 479 <span class="s">}</span>
|
|
493 480 <span class="s">}</span>
|
|
494 481 <span class="s">}</span>
|
|
495 482
|
|
496 483 <span class="c"># Write out the matrices...</span>
|
|
497 484 <span class="k">for</span> <span class="i">$ColNum1</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
498 485 <span class="i">@CorrelationRowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
499 486 <span class="i">@RSquareRowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
500 487 <span class="i">@CovarianceRowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
501 488 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
502 489 <span class="k">push</span> <span class="i">@CorrelationRowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum1</span>]<span class="sc">;</span>
|
|
503 490 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
504 491 <span class="k">push</span> <span class="i">@RSquareRowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum1</span>]<span class="sc">;</span>
|
|
505 492 <span class="s">}</span>
|
|
506 493 <span class="s">}</span>
|
|
507 494 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
508 495 <span class="k">push</span> <span class="i">@CovarianceRowValues</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum1</span>]<span class="sc">;</span>
|
|
509 496 <span class="s">}</span>
|
|
510 497 <span class="k">for</span> <span class="i">$ColNum2</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
511 498 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
512 499 <span class="k">push</span> <span class="i">@CorrelationRowValues</span><span class="cm">,</span> <span class="i">$CorrelationMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>}<span class="sc">;</span>
|
|
513 500 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
514 501 <span class="k">push</span> <span class="i">@RSquareRowValues</span><span class="cm">,</span> <span class="i">$RSquareMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>}<span class="sc">;</span>
|
|
515 502 <span class="s">}</span>
|
|
516 503 <span class="s">}</span>
|
|
517 504 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
518 505 <span class="k">push</span> <span class="i">@CovarianceRowValues</span><span class="cm">,</span> <span class="i">$CovarianceMatrixMap</span>{<span class="i">$ColNum1</span>}{<span class="i">$ColNum2</span>}<span class="sc">;</span>
|
|
519 506 <span class="s">}</span>
|
|
520 507 <span class="s">}</span>
|
|
521 508 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
522 509 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@CorrelationRowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
523 510 <span class="k">print</span> <span class="i">CORRELATIONTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
524 511 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
525 512 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@RSquareRowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
526 513 <span class="k">print</span> <span class="i">RSQUARETEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
527 514 <span class="s">}</span>
|
|
528 515 <span class="s">}</span>
|
|
529 516 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
530 517 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@CovarianceRowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
531 518 <span class="k">print</span> <span class="i">COVARIANCETEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
532 519 <span class="s">}</span>
|
|
533 520 <span class="s">}</span>
|
|
534 521 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCorrelation</span> || <span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
535 522 <span class="k">close</span> <span class="w">CORRELATIONTEXTFILE</span><span class="sc">;</span>
|
|
536 523 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateRSquare</span><span class="s">)</span> <span class="s">{</span>
|
|
537 524 <span class="k">close</span> <span class="w">RSQUARETEXTFILE</span><span class="sc">;</span>
|
|
538 525 <span class="s">}</span>
|
|
539 526 <span class="s">}</span>
|
|
540 527 <span class="k">if</span> <span class="s">(</span><span class="i">$CalculateCovariance</span><span class="s">)</span> <span class="s">{</span>
|
|
541 528 <span class="k">close</span> <span class="w">COVARIANCETEXTFILE</span><span class="sc">;</span>
|
|
542 529 <span class="s">}</span>
|
|
543 530 <span class="s">}</span>
|
|
544 531
|
|
545 532 <span class="c"># Calculate standard scores...</span>
|
|
546 <a name="PerformStandardScoresAnalysis-"></a> 533 <span class="k">sub </span><span class="m">PerformStandardScoresAnalysis</span> <span class="s">{</span>
|
|
547 534 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$ColValuesToAnalyzeMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
548 535 <span class="k">my</span><span class="s">(</span><span class="i">$StandardScores</span><span class="cm">,</span> <span class="i">$StandardScoresN</span><span class="cm">,</span> <span class="i">$NewTextFile</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$Label</span><span class="cm">,</span> <span class="i">$NewLine</span><span class="s">)</span><span class="sc">;</span>
|
|
549 536
|
|
550 537 <span class="i">$StandardScores</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">standardscores</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
551 538 <span class="i">$StandardScoresN</span> = <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">standardscoresn</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
552 539
|
|
553 540 <span class="i">$NewTextFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="q">"StandardScores."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
554 541 <span class="k">print</span> <span class="q">"Generating new text file $NewTextFile...\n"</span><span class="sc">;</span>
|
|
555 542 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$NewTextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $NewTextFile: $! \n"</span><span class="sc">;</span>
|
|
556 543
|
|
557 544 <span class="k">my</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="s">)</span><span class="sc">;</span>
|
|
558 545 <span class="c"># Write out column labels...</span>
|
|
559 546 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
560 547 <span class="i">@ColNumsToAnalyze</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
561 548 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
562 549 <span class="i">$Label</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
563 550 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScores</span><span class="s">)</span> <span class="s">{</span>
|
|
564 551 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"${Label}\(StandardScores)"</span><span class="sc">;</span>
|
|
565 552 <span class="s">}</span>
|
|
566 553 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScoresN</span><span class="s">)</span> <span class="s">{</span>
|
|
567 554 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">"${Label}\(StandardScoresN)"</span><span class="sc">;</span>
|
|
568 555 <span class="s">}</span>
|
|
569 556 <span class="s">}</span>
|
|
570 557 <span class="i">$NewLine</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
571 558 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$NewLine\n"</span><span class="sc">;</span>
|
|
572 559
|
|
573 560 <span class="c"># Go over each column to be analyzed and calculate standard deviation</span>
|
|
574 561 <span class="c"># and mean values...</span>
|
|
575 562 <span class="k">my</span><span class="s">(</span><span class="i">%StandardDeviationMap</span><span class="cm">,</span> <span class="i">%StandardDeviationNMap</span><span class="cm">,</span> <span class="i">%MeanMap</span><span class="s">)</span><span class="sc">;</span>
|
|
576 563 <span class="i">%StandardDeviationMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
577 564 <span class="i">%StandardDeviationNMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
578 565 <span class="i">%MeanMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
579 566 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
580 567 <span class="i">$ColValuesRef</span> = \<span class="i">@</span>{<span class="i">$ColValuesToAnalyzeMapRef</span>->{<span class="i">$ColNum</span>}}<span class="sc">;</span>
|
|
581 568 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$MeanMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
582 569 <span class="i">$MeanMap</span>{<span class="i">$ColNum</span>} = <span class="i">Mean</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
583 570 <span class="s">}</span>
|
|
584 571 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScores</span><span class="s">)</span> <span class="s">{</span>
|
|
585 572 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$StandardDeviationMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
586 573 <span class="i">$StandardDeviationMap</span>{<span class="i">$ColNum</span>} = <span class="i">StandardDeviation</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
587 574 <span class="s">}</span>
|
|
588 575 <span class="s">}</span>
|
|
589 576 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScoresN</span><span class="s">)</span> <span class="s">{</span>
|
|
590 577 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$StandardDeviationNMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
591 578 <span class="i">$StandardDeviationNMap</span>{<span class="i">$ColNum</span>} = <span class="i">StandardDeviationN</span><span class="s">(</span><span class="i">$ColValuesRef</span><span class="s">)</span><span class="sc">;</span>
|
|
592 579 <span class="s">}</span>
|
|
593 580 <span class="s">}</span>
|
|
594 581 <span class="s">}</span>
|
|
595 582 <span class="c">#</span>
|
|
596 583 <span class="c"># Go over each row and calculate standard scores for each column</span>
|
|
597 584 <span class="c"># using (x[i] - mean) / (n - 1) for StandardScores and (x[i] - mean) / n</span>
|
|
598 585 <span class="c"># for StandardScoresN; write out the calculated values as well...</span>
|
|
599 586
|
|
600 587 <span class="k">my</span><span class="s">(</span><span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$ValueOkay</span><span class="cm">,</span> <span class="i">$ScoreValue</span><span class="cm">,</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="s">)</span><span class="sc">;</span>
|
|
601 588 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
602 589 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
603 590
|
|
604 591 <span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $TextFile: $! \n"</span><span class="sc">;</span>
|
|
605 592 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
606 593 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
607 594 <span class="i">@LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
608 595 <span class="i">@RowValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
609 596 <span class="j">COLNUM:</span> <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
610 597 <span class="i">$Value</span> = <span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
611 598 <span class="i">$ValueOkay</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>} && !<span class="i">IsNumerical</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> ? <span class="n">0</span> <span class="co">:</span> <span class="n">1</span><span class="sc">;</span>
|
|
612 599 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScores</span><span class="s">)</span> <span class="s">{</span>
|
|
613 600 <span class="i">$ScoreValue</span> = <span class="i">$ValueOkay</span> ? <span class="s">(</span><span class="s">(</span><span class="i">$Value</span> - <span class="i">$MeanMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span>/<span class="i">$StandardDeviationMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
614 601 <span class="i">$ScoreValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$ScoreValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$ScoreValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$ScoreValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
615 602 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$ScoreValue</span><span class="sc">;</span>
|
|
616 603 <span class="s">}</span>
|
|
617 604 <span class="k">if</span> <span class="s">(</span><span class="i">$StandardScoresN</span><span class="s">)</span> <span class="s">{</span>
|
|
618 605 <span class="i">$ScoreValue</span> = <span class="i">$ValueOkay</span> ? <span class="s">(</span><span class="s">(</span><span class="i">$Value</span> - <span class="i">$MeanMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span>/<span class="i">$StandardDeviationNMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
619 606 <span class="i">$ScoreValue</span> = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$ScoreValue</span><span class="s">)</span> && <span class="k">length</span><span class="s">(</span><span class="i">$ScoreValue</span><span class="s">)</span><span class="s">)</span> ? <span class="s">(</span><span class="k">sprintf</span><span class="s">(</span><span class="q">"%.$OptionsInfo{Precision}f"</span><span class="cm">,</span> <span class="i">$ScoreValue</span><span class="s">)</span> + <span class="n">0</span><span class="s">)</span> <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
620 607 <span class="k">push</span> <span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$ScoreValue</span><span class="sc">;</span>
|
|
621 608 <span class="s">}</span>
|
|
622 609 <span class="s">}</span>
|
|
623 610 <span class="i">$NewLine</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@RowValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
624 611 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$NewLine\n"</span><span class="sc">;</span>
|
|
625 612 <span class="s">}</span>
|
|
626 613 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span>
|
|
627 614 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
628 615 <span class="s">}</span>
|
|
629 616
|
|
630 617 <span class="c"># Make sure the specified columns exists in text files...</span>
|
|
631 <a name="ProcessColumnsInfo-"></a> 618 <span class="k">sub </span><span class="m">ProcessColumnsInfo</span> <span class="s">{</span>
|
|
632 619 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$NewColNum</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">%UniqueColNumsToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
633 620
|
|
634 621 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
635 622 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs1ToAnalyze</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
636 623 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs2ToAnalyze</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
637 624 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">UniqueColNumsToAnalyze</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
638 625
|
|
639 626 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
640 627 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
641 628
|
|
642 629 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
643 630 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs1ToAnalyze</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
644 631 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs2ToAnalyze</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
645 632 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">UniqueColNumsToAnalyze</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
646 633
|
|
647 634 <span class="i">%UniqueColNumsToAnalyzeMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
648 635
|
|
649 636 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
650 637 <span class="i">@ColNumsToAnalyze</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
651 638 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumns</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
652 639 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ColMode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
653 640 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumns</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
654 641 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> >=<span class="n">1</span> && <span class="i">$ColNum</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
655 642 <span class="i">$NewColNum</span> = <span class="i">$ColNum</span> -<span class="n">1</span><span class="sc">;</span>
|
|
656 643 <span class="k">push</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">$NewColNum</span><span class="sc">;</span>
|
|
657 644 <span class="s">}</span>
|
|
658 645 <span class="s">}</span>
|
|
659 646 <span class="s">}</span>
|
|
660 647 <span class="k">else</span> <span class="s">{</span>
|
|
661 648 <span class="k">my</span><span class="s">(</span><span class="i">$ColLabel</span><span class="s">)</span><span class="sc">;</span>
|
|
662 649 <span class="k">for</span> <span class="i">$ColLabel</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumns</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
663 650 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
664 651 <span class="k">push</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="sc">;</span>
|
|
665 652 <span class="s">}</span>
|
|
666 653 <span class="s">}</span>
|
|
667 654 <span class="s">}</span>
|
|
668 655 <span class="s">}</span>
|
|
669 656 <span class="k">elsif</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$OptionsInfo</span>{<span class="w">Columns</span>} && <span class="i">$OptionsInfo</span>{<span class="w">Columns</span>} =~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
670 657 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
671 658 <span class="k">push</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
672 659 <span class="s">}</span>
|
|
673 660 <span class="s">}</span>
|
|
674 661 <span class="k">else</span> <span class="s">{</span>
|
|
675 662 <span class="k">push</span> <span class="i">@ColNumsToAnalyze</span><span class="cm">,</span> <span class="n">0</span><span class="sc">;</span>
|
|
676 663 <span class="s">}</span>
|
|
677 664 <span class="k">if</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
678 665 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@ColNumsToAnalyze</span><span class="sc">;</span>
|
|
679 666 <span class="c"># Set up unique columns map as well...</span>
|
|
680 667 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
681 668 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$UniqueColNumsToAnalyzeMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
682 669 <span class="i">$UniqueColNumsToAnalyzeMap</span>{<span class="i">$ColNum</span>} = <span class="i">$ColNum</span><span class="sc">;</span>
|
|
683 670 <span class="s">}</span>
|
|
684 671 <span class="s">}</span>
|
|
685 672 <span class="s">}</span>
|
|
686 673 <span class="k">else</span> <span class="s">{</span>
|
|
687 674 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: None of the columns specified, @{$OptionsInfo{SpecifiedColumns}}, using \"--columns\" option exist.\n"</span><span class="sc">;</span>
|
|
688 675 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
689 676 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
690 677 <span class="s">}</span>
|
|
691 678 <span class="k">if</span> <span class="s">(</span>!<span class="i">$OptionsInfo</span>{<span class="w">Overwrite</span>} && <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">frequency</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
692 679 <span class="c"># Make sure specific frequency files don't exist...</span>
|
|
693 680 <span class="k">my</span><span class="s">(</span><span class="i">$FrequencyFile</span><span class="s">)</span><span class="sc">;</span>
|
|
694 681 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColNumsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
695 682 <span class="i">$FrequencyFile</span> = <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] . <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>] . <span class="q">"FrequencyAnalysis."</span> . <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
696 683 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$FrequencyFile</span><span class="s">)</span> <span class="s">{</span>
|
|
697 684 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file $FrequencyFile already exists.\n"</span><span class="sc">;</span>
|
|
698 685 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
699 686 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
700 687 <span class="s">}</span>
|
|
701 688 <span class="s">}</span>
|
|
702 689 <span class="s">}</span>
|
|
703 690 <span class="c"># Setup specified column pairs...</span>
|
|
704 691 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>} || <span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>} || <span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span> <span class="s">{</span>
|
|
705 692 <span class="k">my</span><span class="s">(</span><span class="i">@ColPairsToAnalyze</span><span class="cm">,</span> <span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="s">)</span><span class="sc">;</span>
|
|
706 693 <span class="k">if</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
707 694 <span class="c"># Make sure both columns exist...</span>
|
|
708 695 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ColMode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
709 696 <span class="k">for</span> <span class="s">(</span><span class="i">$ColIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="s">(</span><span class="s">(</span><span class="i">$ColIndex</span> + <span class="n">1</span><span class="s">)</span> < <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}}<span class="s">)</span><span class="sc">;</span> <span class="i">$ColIndex</span> += <span class="n">2</span> <span class="s">)</span> <span class="s">{</span>
|
|
710 697 <span class="i">$ColNum1</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
711 698 <span class="i">$ColNum2</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}[<span class="i">$ColIndex</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
712 699 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum1</span> >=<span class="n">1</span> && <span class="i">$ColNum1</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] && <span class="i">$ColNum2</span> >=<span class="n">1</span> && <span class="i">$ColNum2</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
713 700 <span class="i">$ColNum1</span> -= <span class="n">1</span><span class="sc">;</span>
|
|
714 701 <span class="i">$ColNum2</span> -= <span class="n">1</span><span class="sc">;</span>
|
|
715 702 <span class="k">push</span> <span class="i">@ColPairsToAnalyze</span><span class="cm">,</span> <span class="s">(</span><span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="s">)</span><span class="sc">;</span>
|
|
716 703 <span class="s">}</span>
|
|
717 704 <span class="s">}</span>
|
|
718 705 <span class="s">}</span>
|
|
719 706 <span class="k">else</span> <span class="s">{</span>
|
|
720 707 <span class="k">my</span><span class="s">(</span><span class="i">$ColLabel1</span><span class="cm">,</span> <span class="i">$ColLabel2</span><span class="s">)</span><span class="sc">;</span>
|
|
721 708 <span class="k">for</span> <span class="s">(</span><span class="i">$ColIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="s">(</span><span class="s">(</span><span class="i">$ColIndex</span> + <span class="n">1</span><span class="s">)</span> < <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}}<span class="s">)</span><span class="sc">;</span> <span class="i">$ColIndex</span> += <span class="n">2</span> <span class="s">)</span> <span class="s">{</span>
|
|
722 709 <span class="i">$ColLabel1</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
723 710 <span class="i">$ColLabel2</span> = <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}[<span class="i">$ColIndex</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
724 711 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel1</span>}<span class="s">)</span> && <span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel2</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
725 712 <span class="i">$ColNum1</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel1</span>}<span class="sc">;</span>
|
|
726 713 <span class="i">$ColNum2</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel2</span>}<span class="sc">;</span>
|
|
727 714 <span class="k">push</span> <span class="i">@ColPairsToAnalyze</span><span class="cm">,</span> <span class="s">(</span><span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="s">)</span><span class="sc">;</span>
|
|
728 715 <span class="s">}</span>
|
|
729 716 <span class="s">}</span>
|
|
730 717 <span class="s">}</span>
|
|
731 718 <span class="s">}</span>
|
|
732 719 <span class="k">elsif</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AllColumnPairs</span>}<span class="s">)</span> <span class="s">{</span>
|
|
733 720 <span class="k">for</span> <span class="i">$ColNum1</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
734 721 <span class="k">for</span> <span class="i">$ColNum2</span> <span class="s">(</span><span class="n">0</span> .. <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] - <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
735 722 <span class="k">push</span> <span class="i">@ColPairsToAnalyze</span><span class="cm">,</span> <span class="s">(</span><span class="i">$ColNum1</span><span class="cm">,</span> <span class="i">$ColNum2</span><span class="s">)</span><span class="sc">;</span>
|
|
736 723 <span class="s">}</span>
|
|
737 724 <span class="s">}</span>
|
|
738 725 <span class="s">}</span>
|
|
739 726 <span class="k">else</span> <span class="s">{</span>
|
|
740 727 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] >= <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
741 728 <span class="k">push</span> <span class="i">@ColPairsToAnalyze</span><span class="cm">,</span> <span class="s">(</span><span class="n">0</span><span class="cm">,</span><span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
742 729 <span class="s">}</span>
|
|
743 730 <span class="s">}</span>
|
|
744 731 <span class="k">if</span> <span class="s">(</span><span class="i">@ColPairsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
745 732 <span class="k">if</span> <span class="s">(</span><span class="i">@ColPairsToAnalyze</span> % <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
746 733 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: Invalid number values specified using \"--columnpairs\" option: It must contain even number of valid values.\n"</span><span class="sc">;</span>
|
|
747 734 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
748 735 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
749 736 <span class="s">}</span>
|
|
750 737 <span class="k">else</span> <span class="s">{</span>
|
|
751 738 <span class="k">for</span> <span class="s">(</span><span class="i">$ColIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$ColIndex</span> < <span class="i">@ColPairsToAnalyze</span><span class="sc">;</span> <span class="i">$ColIndex</span> += <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
752 739 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs1ToAnalyze</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">$ColPairsToAnalyze</span>[<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
753 740 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColPairs2ToAnalyze</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">$ColPairsToAnalyze</span>[<span class="i">$ColIndex</span> + <span class="n">1</span>]<span class="sc">;</span>
|
|
754 741 <span class="s">}</span>
|
|
755 742 <span class="c"># Set up unique columns map as well...</span>
|
|
756 743 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@ColPairsToAnalyze</span><span class="s">)</span> <span class="s">{</span>
|
|
757 744 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span> <span class="i">$UniqueColNumsToAnalyzeMap</span>{<span class="i">$ColNum</span>}<span class="s">)</span> <span class="s">{</span>
|
|
758 745 <span class="i">$UniqueColNumsToAnalyzeMap</span>{<span class="i">$ColNum</span>} = <span class="i">$ColNum</span><span class="sc">;</span>
|
|
759 746 <span class="s">}</span>
|
|
760 747 <span class="s">}</span>
|
|
761 748 <span class="s">}</span>
|
|
762 749 <span class="s">}</span>
|
|
763 750 <span class="s">}</span>
|
|
764 751 <span class="c"># Setup uniques columns array...</span>
|
|
765 752 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">UniqueColNumsToAnalyze</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="s">(</span><span class="k">sort</span> <span class="k">keys</span> <span class="i">%UniqueColNumsToAnalyzeMap</span><span class="s">)</span><span class="sc">;</span>
|
|
766 753 <span class="s">}</span>
|
|
767 754 <span class="s">}</span>
|
|
768 755 <span class="s">}</span>
|
|
769 756
|
|
770 757 <span class="c"># Retrieve information about input text files...</span>
|
|
771 <a name="RetrieveTextFilesInfo-"></a> 758 <span class="k">sub </span><span class="m">RetrieveTextFilesInfo</span> <span class="s">{</span>
|
|
772 759 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OutFileRoot</span><span class="cm">,</span> <span class="i">$OutFile</span><span class="cm">,</span> <span class="i">$OutFileExt</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="s">)</span><span class="sc">;</span>
|
|
773 760
|
|
774 761 <span class="i">%TextFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
775 762
|
|
776 763 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
777 764 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
778 765 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
779 766 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
780 767 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
781 768 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
782 769 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
783 770
|
|
784 771 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
785 772 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
786 773
|
|
787 774 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
788 775 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
789 776 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
790 777 <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
791 778 <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
792 779
|
|
793 780 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
794 781 <span class="i">%</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
795 782
|
|
796 783 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$TextFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
797 784 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It doesn't exist\n"</span><span class="sc">;</span>
|
|
798 785 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
799 786 <span class="s">}</span>
|
|
800 787 <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$TextFile</span><span class="cm">,</span> <span class="q">"csv tsv"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
801 788 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It's not a csv or tsv file\n"</span><span class="sc">;</span>
|
|
802 789 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
803 790 <span class="s">}</span>
|
|
804 791 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$TextFile</span><span class="s">)</span><span class="sc">;</span>
|
|
805 792 <span class="k">if</span> <span class="s">(</span><span class="i">$FileExt</span> =~ <span class="q">/^tsv$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
806 793 <span class="i">$InDelim</span> = <span class="q">"\t"</span><span class="sc">;</span>
|
|
807 794 <span class="s">}</span>
|
|
808 795 <span class="k">else</span> <span class="s">{</span>
|
|
809 796 <span class="i">$InDelim</span> = <span class="q">"\,"</span><span class="sc">;</span>
|
|
810 797 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
811 798 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The value specified, $Options{indelim}, for option \"--indelim\" is not valid for csv files\n"</span><span class="sc">;</span>
|
|
812 799 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
813 800 <span class="s">}</span>
|
|
814 801 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} =~ <span class="q">/^semicolon$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
815 802 <span class="i">$InDelim</span> = <span class="q">"\;"</span><span class="sc">;</span>
|
|
816 803 <span class="s">}</span>
|
|
817 804 <span class="s">}</span>
|
|
818 805
|
|
819 806 <span class="k">if</span> <span class="s">(</span>!<span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span><span class="s">)</span> <span class="s">{</span>
|
|
820 807 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: Couldn't open it: $! \n"</span><span class="sc">;</span>
|
|
821 808 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
822 809 <span class="s">}</span>
|
|
823 810
|
|
824 811 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
825 812 <span class="i">@ColLabels</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
826 813 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span>
|
|
827 814
|
|
828 815 <span class="i">$FileDir</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileName</span> = <span class="q">""</span><span class="sc">;</span> <span class="i">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
829 816 <span class="s">(</span><span class="i">$FileDir</span><span class="cm">,</span> <span class="i">$FileName</span><span class="cm">,</span> <span class="i">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$TextFile</span><span class="s">)</span><span class="sc">;</span>
|
|
830 817 <span class="i">$FileExt</span> = <span class="q">"csv"</span><span class="sc">;</span>
|
|
831 818 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
832 819 <span class="i">$FileExt</span> = <span class="q">"tsv"</span><span class="sc">;</span>
|
|
833 820 <span class="s">}</span>
|
|
834 821 <span class="i">$OutFileExt</span> = <span class="i">$FileExt</span><span class="sc">;</span>
|
|
835 822 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">root</span>} && <span class="s">(</span><span class="i">@TextFilesList</span> == <span class="n">1</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
836 823 <span class="k">my</span> <span class="s">(</span><span class="i">$RootFileDir</span><span class="cm">,</span> <span class="i">$RootFileName</span><span class="cm">,</span> <span class="i">$RootFileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">root</span>}<span class="s">)</span><span class="sc">;</span>
|
|
837 824 <span class="k">if</span> <span class="s">(</span><span class="i">$RootFileName</span> && <span class="i">$RootFileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
838 825 <span class="i">$FileName</span> = <span class="i">$RootFileName</span><span class="sc">;</span>
|
|
839 826 <span class="s">}</span>
|
|
840 827 <span class="k">else</span> <span class="s">{</span>
|
|
841 828 <span class="i">$FileName</span> = <span class="i">$Options</span>{<span class="w">root</span>}<span class="sc">;</span>
|
|
842 829 <span class="s">}</span>
|
|
843 830 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
844 831 <span class="s">}</span>
|
|
845 832 <span class="k">else</span> <span class="s">{</span>
|
|
846 833 <span class="i">$OutFileRoot</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
847 834 <span class="s">}</span>
|
|
848 835 <span class="i">$OutFile</span> = <span class="i">$OutFileRoot</span> . <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} . <span class="q">".$OutFileExt"</span><span class="sc">;</span>
|
|
849 836
|
|
850 837 <span class="k">if</span> <span class="s">(</span><span class="k">lc</span><span class="s">(</span><span class="i">$OutFile</span><span class="s">)</span> <span class="k">eq</span> <span class="k">lc</span><span class="s">(</span><span class="i">$TextFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
851 838 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile:Output file name, $OutFile, is same as input text file name, $TextFile\n"</span><span class="sc">;</span>
|
|
852 839 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
853 840 <span class="s">}</span>
|
|
854 841 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">overwrite</span>}<span class="s">)</span> <span class="s">{</span>
|
|
855 842 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$OutFile</span><span class="s">)</span> <span class="s">{</span>
|
|
856 843 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file $OutFile already exists\n"</span><span class="sc">;</span>
|
|
857 844 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
858 845 <span class="s">}</span>
|
|
859 846 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>}<span class="s">)</span> || <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>}<span class="s">)</span> || <span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
860 847 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">AllColumnPairs</span>}<span class="s">)</span> <span class="s">{</span>
|
|
861 848 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">covariance</span>}<span class="s">)</span> && <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}CovarianceMatrix.${FileExt}"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
862 849 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file ${OutFileRoot}Covariance.${FileExt} already exists.\n"</span><span class="sc">;</span>
|
|
863 850 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
864 851 <span class="s">}</span>
|
|
865 852 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">correlation</span>}<span class="s">)</span> && <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}CorrelationMatrix.${FileExt}"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
866 853 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file ${OutFileRoot}CorrelationMatrix.${FileExt} already exists.\n"</span><span class="sc">;</span>
|
|
867 854 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
868 855 <span class="s">}</span>
|
|
869 856 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">rsquare</span>}<span class="s">)</span> && <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}RSquareMatrix.${FileExt}"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
870 857 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file ${OutFileRoot}RSquareMatrix.${FileExt} already exists.\n"</span><span class="sc">;</span>
|
|
871 858 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
872 859 <span class="s">}</span>
|
|
873 860 <span class="s">}</span>
|
|
874 861 <span class="k">else</span> <span class="s">{</span>
|
|
875 862 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}ColumnPairsAnalysis.${FileExt}"</span><span class="s">)</span> <span class="s">{</span>
|
|
876 863 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file ${OutFileRoot}ColumnPairsAnalysis.${FileExt} already exists.\n"</span><span class="sc">;</span>
|
|
877 864 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
878 865 <span class="s">}</span>
|
|
879 866 <span class="s">}</span>
|
|
880 867 <span class="s">}</span>
|
|
881 868 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="w">standardscores</span>}<span class="s">)</span> && <span class="s">(</span><span class="k">-e</span> <span class="q">"${OutFileRoot}StandardScores.${FileExt}"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
882 869 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The file ${OutFileRoot}StandardScores.${FileExt} already exists.\n"</span><span class="sc">;</span>
|
|
883 870 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
884 871 <span class="s">}</span>
|
|
885 872 <span class="s">}</span>
|
|
886 873
|
|
887 874 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
888 875 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="i">$InDelim</span><span class="sc">;</span>
|
|
889 876 <span class="i">$TextFilesInfo</span>{<span class="w">OutFileRoot</span>}[<span class="i">$Index</span>] = <span class="q">"$OutFileRoot"</span><span class="sc">;</span>
|
|
890 877 <span class="i">$TextFilesInfo</span>{<span class="w">OutFileExt</span>}[<span class="i">$Index</span>] = <span class="q">"$OutFileExt"</span><span class="sc">;</span>
|
|
891 878
|
|
892 879 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="i">@ColLabels</span><span class="sc">;</span>
|
|
893 880 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@ColLabels</span><span class="sc">;</span>
|
|
894 881 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#ColLabels</span><span class="s">)</span> <span class="s">{</span>
|
|
895 882 <span class="i">$ColLabel</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
896 883 <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>} = <span class="i">$ColNum</span><span class="sc">;</span>
|
|
897 884 <span class="s">}</span>
|
|
898 885 <span class="s">}</span>
|
|
899 886 <span class="s">}</span>
|
|
900 887
|
|
901 888 <span class="c"># Process option values...</span>
|
|
902 <a name="ProcessOptions-"></a> 889 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
903 890 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
904 891
|
|
905 892 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
906 893
|
|
907 894 <span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} = <span class="i">$Options</span>{<span class="w">detail</span>}<span class="sc">;</span>
|
|
908 895
|
|
909 896 <span class="c"># Setup supported statistical functions...</span>
|
|
910 897 <span class="k">my</span><span class="s">(</span><span class="i">$SupportedFunction</span><span class="cm">,</span> <span class="i">@SupportedStatisticaFunctions</span><span class="cm">,</span> <span class="i">%SupportedStatisticaFunctionsMap</span><span class="s">)</span><span class="sc">;</span>
|
|
911 898 <span class="i">%SupportedStatisticaFunctionsMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
912 899 <span class="i">@SupportedStatisticaFunctions</span> = <span class="q">qw(Average AverageDeviation Correlation Count Covariance GeometricMean Frequency HarmonicMean KLargest KSmallest Kurtosis Maximum Minimum Mean Median Mode RSquare Skewness Sum SumOfSquares StandardDeviation StandardDeviationN StandardError StandardScores StandardScoresN TrimMean Variance VarianceN)</span><span class="sc">;</span>
|
|
913 900
|
|
914 901 <span class="k">for</span> <span class="i">$SupportedFunction</span> <span class="s">(</span><span class="i">@SupportedStatisticaFunctions</span><span class="s">)</span> <span class="s">{</span>
|
|
915 902 <span class="i">$SupportedStatisticaFunctionsMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SupportedFunction</span><span class="s">)</span>} = <span class="i">$SupportedFunction</span><span class="sc">;</span>
|
|
916 903 <span class="s">}</span>
|
|
917 904
|
|
918 905 <span class="c"># Setup a list of functions to use for analysis...</span>
|
|
919 906 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span><span class="sc">;</span>
|
|
920 907 <span class="i">%</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
921 908 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
922 909 <span class="c"># Check mode values...</span>
|
|
923 910 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^DescriptiveStatisticsBasic$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
924 911 <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"DescriptiveStatisticsBasic"</span><span class="sc">;</span>
|
|
925 912 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}} = <span class="q">qw(Count Maximum Minimum Mean Median StandardDeviation StandardError Variance Sum)</span><span class="sc">;</span>
|
|
926 913 <span class="s">}</span>
|
|
927 914 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^DescriptiveStatisticsAll$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
928 915 <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"DescriptiveStatisticsAll"</span><span class="sc">;</span>
|
|
929 916 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}} = <span class="q">qw(Count Maximum Minimum Mean GeometricMean HarmonicMean TrimMean Median Mode StandardDeviation Kurtosis Skewness StandardError Variance RSquare Frequency KLargest KSmallest Sum)</span><span class="sc">;</span>
|
|
930 917 <span class="s">}</span>
|
|
931 918 <span class="k">elsif</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^All$/i</span> <span class="s">)</span> <span class="s">{</span>
|
|
932 919 <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"AllStatistics"</span><span class="sc">;</span>
|
|
933 920 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}} = <span class="i">@SupportedStatisticaFunctions</span><span class="sc">;</span>
|
|
934 921 <span class="s">}</span>
|
|
935 922 <span class="k">else</span> <span class="s">{</span>
|
|
936 923 <span class="i">$OptionsInfo</span>{<span class="w">FileNameMode</span>} = <span class="q">"SpecifiedStatistics"</span><span class="sc">;</span>
|
|
937 924 <span class="c"># Comma delimited list of functions...</span>
|
|
938 925 <span class="k">my</span><span class="s">(</span><span class="i">$Mode</span><span class="cm">,</span> <span class="i">@SpecifiedFunctions</span><span class="cm">,</span> <span class="i">@UnsupportedSpecifiedFunctions</span><span class="s">)</span><span class="sc">;</span>
|
|
939 926 <span class="i">$Mode</span> = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
940 927 <span class="i">$Mode</span> =~ <span class="q">s/ //g</span><span class="sc">;</span>
|
|
941 928 <span class="i">@SpecifiedFunctions</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Mode</span><span class="sc">;</span>
|
|
942 929 <span class="i">@UnsupportedSpecifiedFunctions</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
943 930 <span class="k">for</span> <span class="i">$SpecifiedFunction</span> <span class="s">(</span><span class="i">@SpecifiedFunctions</span><span class="s">)</span> <span class="s">{</span>
|
|
944 931 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$SupportedStatisticaFunctionsMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
945 932 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}}<span class="cm">,</span> <span class="i">$SpecifiedFunction</span><span class="sc">;</span>
|
|
946 933 <span class="s">}</span>
|
|
947 934 <span class="k">else</span> <span class="s">{</span>
|
|
948 935 <span class="k">push</span> <span class="i">@UnsupportedSpecifiedFunctions</span><span class="cm">,</span> <span class="i">$SpecifiedFunction</span><span class="sc">;</span>
|
|
949 936 <span class="s">}</span>
|
|
950 937 <span class="s">}</span>
|
|
951 938 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedFunctions</span><span class="s">)</span> <span class="s">{</span>
|
|
952 939 <span class="k">if</span> <span class="s">(</span><span class="i">@UnsupportedSpecifiedFunctions</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
953 940 <span class="k">warn</span> <span class="q">"Error: The values specified - "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@UnsupportedSpecifiedFunctions</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span><span class="q">" - for option \"-m --mode\" are not valid.\n"</span><span class="sc">;</span>
|
|
954 941 <span class="s">}</span>
|
|
955 942 <span class="k">else</span> <span class="s">{</span>
|
|
956 943 <span class="k">warn</span> <span class="q">"Error: The value specified, @UnsupportedSpecifiedFunctions , for option \"-m --mode\" is not valid.\n"</span><span class="sc">;</span>
|
|
957 944 <span class="s">}</span>
|
|
958 945 <span class="k">die</span> <span class="q">"Allowed values:"</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@SupportedStatisticaFunctions</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">0</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
959 946 <span class="s">}</span>
|
|
960 947 <span class="s">}</span>
|
|
961 948 <span class="j">FUNCTION:</span> <span class="k">for</span> <span class="i">$SpecifiedFunction</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctions</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
962 949 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span> <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span>} <span class="s">)</span> <span class="s">{</span>
|
|
963 950 <span class="k">next</span> <span class="j">FUNCTION</span><span class="sc">;</span>
|
|
964 951 <span class="s">}</span>
|
|
965 952 <span class="i">$OptionsInfo</span>{<span class="w">SpecifiedStatisticalFunctionsMap</span>}{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span>} = <span class="i">$SupportedStatisticaFunctionsMap</span>{<span class="k">lc</span><span class="s">(</span><span class="i">$SpecifiedFunction</span><span class="s">)</span>}<span class="sc">;</span>
|
|
966 953 <span class="s">}</span>
|
|
967 954
|
|
968 955 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/tab/i</span> <span class="s">)</span> ? <span class="q">"\t"</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/semicolon/i</span><span class="s">)</span> ? <span class="q">"\;"</span> <span class="co">:</span> <span class="q">"\,"</span><span class="s">)</span><span class="sc">;</span>
|
|
969 956 <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/yes/i</span> <span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
970 957
|
|
971 958 <span class="i">$OptionsInfo</span>{<span class="w">Overwrite</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="i">$Options</span>{<span class="w">overwrite</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
972 959 <span class="i">$OptionsInfo</span>{<span class="w">Root</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
973 960
|
|
974 961 <span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>} = <span class="i">$Options</span>{<span class="w">fast</span>} ? <span class="n">0</span> <span class="co">:</span> <span class="n">1</span><span class="sc">;</span>
|
|
975 962 <span class="i">$OptionsInfo</span>{<span class="w">Precision</span>} = <span class="i">$Options</span>{<span class="w">precision</span>}<span class="sc">;</span>
|
|
976 963
|
|
977 964 <span class="i">$OptionsInfo</span>{<span class="w">KLargest</span>} = <span class="i">$Options</span>{<span class="w">klargest</span>}<span class="sc">;</span>
|
|
978 965 <span class="i">$OptionsInfo</span>{<span class="w">KSmallest</span>} = <span class="i">$Options</span>{<span class="w">ksmallest</span>}<span class="sc">;</span>
|
|
979 966
|
|
980 967 <span class="i">$OptionsInfo</span>{<span class="w">TrimFraction</span>} = <span class="i">$Options</span>{<span class="w">trimfraction</span>}<span class="sc">;</span>
|
|
981 968
|
|
982 969 <span class="c"># Setup frequency bin values...</span>
|
|
983 970 <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>} = <span class="n">10</span><span class="sc">;</span>
|
|
984 971 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
985 972 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">frequencybins</span>} =~ <span class="q">/\,/</span><span class="s">)</span> <span class="s">{</span>
|
|
986 973 <span class="k">my</span><span class="s">(</span><span class="i">$BinValue</span><span class="cm">,</span> <span class="i">@SpecifiedBinRange</span><span class="s">)</span><span class="sc">;</span>
|
|
987 974 <span class="i">@SpecifiedBinRange</span> = <span class="k">split</span> <span class="q">/\,/</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">frequencybins</span>}<span class="sc">;</span>
|
|
988 975 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedBinRange</span> < <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
989 976 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{frequencybins}, for option \"--frequencybins\" is not valid: Must contain at least two values. \n"</span><span class="sc">;</span>
|
|
990 977 <span class="s">}</span>
|
|
991 978 <span class="k">for</span> <span class="i">$BinValue</span> <span class="s">(</span><span class="i">@SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
|
|
992 979 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsNumerical</span><span class="s">(</span><span class="i">$BinValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
993 980 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{frequencybins}, for option \"--frequencybins\" is not valid: Contains non numeric values. \n"</span><span class="sc">;</span>
|
|
994 981 <span class="s">}</span>
|
|
995 982 <span class="s">}</span>
|
|
996 983 <span class="k">my</span><span class="s">(</span><span class="i">$Index1</span><span class="cm">,</span> <span class="i">$Index2</span><span class="s">)</span><span class="sc">;</span>
|
|
997 984 <span class="k">for</span> <span class="i">$Index1</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
|
|
998 985 <span class="k">for</span> <span class="i">$Index2</span> <span class="s">(</span><span class="s">(</span><span class="i">$Index1</span> + <span class="n">1</span><span class="s">)</span> .. <span class="i">$#SpecifiedBinRange</span><span class="s">)</span> <span class="s">{</span>
|
|
999 986 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedBinRange</span>[<span class="i">$Index1</span>] >= <span class="i">$SpecifiedBinRange</span>[<span class="i">$Index2</span>]<span class="s">)</span> <span class="s">{</span>
|
|
1000 987 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{frequencybins}, for option \"--frequencybins\" is not valid: Must contain values in ascending order. \n"</span><span class="sc">;</span>
|
|
1001 988 <span class="s">}</span>
|
|
1002 989 <span class="s">}</span>
|
|
1003 990 <span class="s">}</span>
|
|
1004 991 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">BinRange</span>}}<span class="cm">,</span> <span class="i">@SpecifiedBinRange</span><span class="sc">;</span>
|
|
1005 992 <span class="s">}</span>
|
|
1006 993 <span class="k">else</span> <span class="s">{</span>
|
|
1007 994 <span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>} = <span class="i">$Options</span>{<span class="w">frequencybins</span>}<span class="sc">;</span>
|
|
1008 995 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">NumOfBins</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1009 996 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{frequencybins}, for option \"--frequencybins\" is not valid. Allowed values: positive integer or \"number,number,[number]...\". \n"</span><span class="sc">;</span>
|
|
1010 997 <span class="s">}</span>
|
|
1011 998 <span class="s">}</span>
|
|
1012 999
|
|
1013 1000 <span class="c"># Setup specified columns...</span>
|
|
1014 1001 <span class="i">$OptionsInfo</span>{<span class="w">ColMode</span>} = <span class="i">$Options</span>{<span class="w">colmode</span>}<span class="sc">;</span>
|
|
1015 1002 <span class="i">$OptionsInfo</span>{<span class="w">Columns</span>} = <span class="k">defined</span> <span class="i">$Options</span>{<span class="w">columns</span>} ? <span class="i">$Options</span>{<span class="w">columns</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
1016 1003
|
|
1017 1004 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumns</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1018 1005 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span> <span class="i">$Options</span>{<span class="w">columns</span>} && <span class="i">$Options</span>{<span class="w">columns</span>} !~ <span class="q">/^All$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1019 1006 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedValues</span><span class="s">)</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">columns</span>}<span class="sc">;</span>
|
|
1020 1007 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1021 1008 <span class="k">my</span><span class="s">(</span><span class="i">$ColValue</span><span class="s">)</span><span class="sc">;</span>
|
|
1022 1009 <span class="k">for</span> <span class="i">$ColValue</span> <span class="s">(</span><span class="i">@SpecifiedValues</span><span class="s">)</span> <span class="s">{</span>
|
|
1023 1010 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$ColValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1024 1011 <span class="k">die</span> <span class="q">"Error: Column value, $ColValue, specified using \"--columns\" is not valid: Allowed integer values: > 0.\n"</span><span class="sc">;</span>
|
|
1025 1012 <span class="s">}</span>
|
|
1026 1013 <span class="s">}</span>
|
|
1027 1014 <span class="s">}</span>
|
|
1028 1015 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumns</span>}}<span class="cm">,</span> <span class="i">@SpecifiedValues</span><span class="sc">;</span>
|
|
1029 1016 <span class="s">}</span>
|
|
1030 1017 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1031 1018 <span class="i">$OptionsInfo</span>{<span class="w">AllColumnPairs</span>} = <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">columnpairs</span>}<span class="s">)</span> && <span class="i">$Options</span>{<span class="w">columnpairs</span>} =~ <span class="q">/^AllPairs$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
1032 1019 <span class="k">if</span> <span class="s">(</span><span class="k">defined</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">columnpairs</span>}<span class="s">)</span> && !<span class="i">$OptionsInfo</span>{<span class="w">AllColumnPairs</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1033 1020 <span class="k">my</span><span class="s">(</span><span class="i">@SpecifiedValues</span><span class="s">)</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">columnpairs</span>}<span class="sc">;</span>
|
|
1034 1021 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedValues</span> % <span class="n">2</span><span class="s">)</span> <span class="s">{</span>
|
|
1035 1022 <span class="k">die</span> <span class="q">"Error: Invalid number of values specified using \"--columnpairs\" option: It must contain even number of values.\n"</span><span class="sc">;</span>
|
|
1036 1023 <span class="s">}</span>
|
|
1037 1024 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1038 1025 <span class="k">my</span><span class="s">(</span><span class="i">$ColValue</span><span class="s">)</span><span class="sc">;</span>
|
|
1039 1026 <span class="k">for</span> <span class="i">$ColValue</span> <span class="s">(</span><span class="i">@SpecifiedValues</span><span class="s">)</span> <span class="s">{</span>
|
|
1040 1027 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$ColValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1041 1028 <span class="k">die</span> <span class="q">"Error: Column value, $ColValue, specified using \"--columnpairs\" is not valid: Allowed integer values: > 0.\n"</span><span class="sc">;</span>
|
|
1042 1029 <span class="s">}</span>
|
|
1043 1030 <span class="s">}</span>
|
|
1044 1031 <span class="s">}</span>
|
|
1045 1032 <span class="k">push</span> <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedColumnPairs</span>}}<span class="cm">,</span> <span class="i">@SpecifiedValues</span><span class="sc">;</span>
|
|
1046 1033 <span class="s">}</span>
|
|
1047 1034
|
|
1048 1035 <span class="s">}</span>
|
|
1049 1036
|
|
1050 1037 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
1051 <a name="SetupScriptUsage-"></a>1038 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
1052 1039
|
|
1053 1040 <span class="c"># Retrieve all the options...</span>
|
|
1054 1041 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
1055 1042 <span class="i">$Options</span>{<span class="w">colmode</span>} = <span class="q">"colnum"</span><span class="sc">;</span>
|
|
1056 1043 <span class="i">$Options</span>{<span class="w">detail</span>} = <span class="n">1</span><span class="sc">;</span>
|
|
1057 1044 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
1058 1045 <span class="i">$Options</span>{<span class="w">frequencybins</span>} = <span class="n">10</span><span class="sc">;</span>
|
|
1059 1046 <span class="i">$Options</span>{<span class="w">klargest</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
1060 1047 <span class="i">$Options</span>{<span class="w">ksmallest</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
1061 1048 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">"DescriptiveStatisticsBasic"</span><span class="sc">;</span>
|
|
1062 1049 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
1063 1050 <span class="i">$Options</span>{<span class="w">precision</span>} = <span class="n">2</span><span class="sc">;</span>
|
|
1064 1051 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">"yes"</span><span class="sc">;</span>
|
|
1065 1052 <span class="i">$Options</span>{<span class="w">trimfraction</span>} = <span class="n">0.1</span><span class="sc">;</span>
|
|
1066 1053
|
|
1067 1054 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">"colmode|c=s"</span><span class="cm">,</span> <span class="q">"columns=s"</span><span class="cm">,</span> <span class="q">"columnpairs=s"</span><span class="cm">,</span> <span class="q">"detail|d=i"</span><span class="cm">,</span> <span class="q">"frequencybins=s"</span><span class="cm">,</span> <span class="q">"fast|f"</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"indelim=s"</span><span class="cm">,</span> <span class="q">"klargest=i"</span><span class="cm">,</span> <span class="q">"ksmallest=i"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"precision|p=i"</span><span class="cm">,</span> <span class="q">"quote|q=s"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"trimfraction=f"</span><span class="cm">,</span> <span class="q">"workingdir|w=s"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1068 1055 <span class="k">die</span> <span class="q">"\nTo get a list of valid options and their values, use \"$ScriptName -h\" or\n\"perl -S $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
1069 1056 <span class="s">}</span>
|
|
1070 1057 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1071 1058 <span class="k">if</span> <span class="s">(</span>! <span class="k">-d</span> <span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
1072 1059 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
1073 1060 <span class="s">}</span>
|
|
1074 1061 <span class="k">chdir</span> <span class="i">$Options</span>{<span class="w">workingdir</span>} <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't chdir $Options{workingdir}: $! \n"</span><span class="sc">;</span>
|
|
1075 1062 <span class="s">}</span>
|
|
1076 1063 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">colmode</span>} !~ <span class="q">/^(colnum|collabel)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1077 1064 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{colmode}, for option \"-c --colmode\" is not valid. Allowed values: colnum or collabel\n"</span><span class="sc">;</span>
|
|
1078 1065 <span class="s">}</span>
|
|
1079 1066 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">detail</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1080 1067 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n"</span><span class="sc">;</span>
|
|
1081 1068 <span class="s">}</span>
|
|
1082 1069 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1083 1070 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n"</span><span class="sc">;</span>
|
|
1084 1071 <span class="s">}</span>
|
|
1085 1072 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1086 1073 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
1087 1074 <span class="s">}</span>
|
|
1088 1075 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
1089 1076 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
1090 1077 <span class="s">}</span>
|
|
1091 1078 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">precision</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1092 1079 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{precision}, for option \"-p --precision\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
1093 1080 <span class="s">}</span>
|
|
1094 1081 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">klargest</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1095 1082 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{klargest}, for option \"--klargest\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
1096 1083 <span class="s">}</span>
|
|
1097 1084 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">ksmallest</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1098 1085 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{ksmallest}, for option \"--ksmallest\" is not valid. Allowed values: > 0 \n"</span><span class="sc">;</span>
|
|
1099 1086 <span class="s">}</span>
|
|
1100 1087 <span class="k">if</span> <span class="s">(</span><span class="i">IsFloat</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">trimfraction</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
1101 1088 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">trimfraction</span>} <= <span class="n">0</span> || <span class="i">$Options</span>{<span class="w">trimfraction</span>} >= <span class="n">1.0</span><span class="s">)</span> <span class="s">{</span>
|
|
1102 1089 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{trimfraction}, for option \"--trimfraction\" is not valid. Allowed values: > 0 and < 1.0\n"</span><span class="sc">;</span>
|
|
1103 1090 <span class="s">}</span>
|
|
1104 1091 <span class="s">}</span>
|
|
1105 1092 <span class="k">else</span> <span class="s">{</span>
|
|
1106 1093 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{trimfraction}, for option \"--trimfraction\" is not valid. Allowed values: > 0 and < 1.0\n"</span><span class="sc">;</span>
|
|
1107 1094 <span class="s">}</span>
|
|
1108 1095 <span class="s">}</span>
|
|
1109 1096
|
|
1110 <a name="EOF-"></a></pre>
|
|
1111 <p> </p>
|
|
1112 <br />
|
|
1113 <center>
|
|
1114 <img src="../../../images/h2o2.png">
|
|
1115 </center>
|
|
1116 </body>
|
|
1117 </html>
|