Mercurial > repos > bgruening > sklearn_regression_metrics
annotate label_encoder.py @ 28:842d9c59b6b7 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
author | bgruening |
---|---|
date | Tue, 20 Apr 2021 16:52:49 +0000 |
parents | |
children | e3e0ae93f7c8 |
rev | line source |
---|---|
28
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
1 import argparse |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
2 import json |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
3 import warnings |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
4 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
5 import numpy as np |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
6 import pandas as pd |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
7 from sklearn.preprocessing import LabelEncoder |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
8 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
9 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
10 def main(inputs, infile, outfile): |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
11 """ |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
12 Parameter |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
13 --------- |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
14 input : str |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
15 File path to galaxy tool parameter |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
16 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
17 infile : str |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
18 File paths of input vector |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
19 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
20 outfile : str |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
21 File path to output vector |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
22 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
23 """ |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
24 warnings.simplefilter('ignore') |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
25 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
26 with open(inputs, 'r') as param_handler: |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
27 params = json.load(param_handler) |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
28 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
29 input_header = params['header0'] |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
30 header = 'infer' if input_header else None |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
31 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
32 input_vector = pd.read_csv(infile, sep='\t', header=header) |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
33 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
34 le = LabelEncoder() |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
35 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
36 output_vector = le.fit_transform(input_vector) |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
37 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
38 np.savetxt(outfile, output_vector, fmt="%d", delimiter='\t') |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
39 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
40 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
41 if __name__ == '__main__': |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
42 aparser = argparse.ArgumentParser() |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
43 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
44 aparser.add_argument("-y", "--infile", dest="infile") |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
45 aparser.add_argument("-o", "--outfile", dest="outfile") |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
46 args = aparser.parse_args() |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
47 |
842d9c59b6b7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
48 main(args.inputs, args.infile, args.outfile) |