Mercurial > repos > bgruening > create_tool_recommendation_model

annotate prepare_data.py @ 5:9ec705bd11cb draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
author bgruening
date Sun, 16 Oct 2022 11:51:32 +0000
parents f0da532be419
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
1 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
2 Prepare the workflow paths to be used by downstream
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
3 machine learning algorithm. The paths are divided
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
4 into the test and training sets
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
5 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
6
4
f0da532be419 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 5eebc0cb44e71f581d548b7e842002705dd155eb"
bgruening
parents: 3
diff changeset
7 import collections
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
8 import random
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
9
4
f0da532be419 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 5eebc0cb44e71f581d548b7e842002705dd155eb"
bgruening
parents: 3
diff changeset
10 import numpy as np
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
11 import predict_tool_usage
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
12 from sklearn.model_selection import train_test_split
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
13
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
14
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
15 class PrepareData:
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
16
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
17 def __init__(self, max_seq_length, test_data_share):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
18 """ Init method. """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
19 self.max_tool_sequence_len = max_seq_length
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
20 self.test_share = test_data_share
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
21
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
22 def process_workflow_paths(self, workflow_paths):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
23 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
24 Get all the tools and complete set of individual paths for each workflow
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
25 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
26 tokens = list()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
27 raw_paths = workflow_paths
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
28 raw_paths = [x.replace("\n", '') for x in raw_paths]
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
29 for item in raw_paths:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
30 split_items = item.split(",")
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
31 for token in split_items:
4
f0da532be419 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 5eebc0cb44e71f581d548b7e842002705dd155eb"
bgruening
parents: 3
diff changeset
32 if token != "":
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
33 tokens.append(token)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
34 tokens = list(set(tokens))
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
35 tokens = np.array(tokens)
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
36 tokens = np.reshape(tokens, [-1, ])
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
37 return tokens, raw_paths
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
38
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
39 def create_new_dict(self, new_data_dict):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
40 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
41 Create new data dictionary
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
42 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
43 reverse_dict = dict((v, k) for k, v in new_data_dict.items())
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
44 return new_data_dict, reverse_dict
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
45
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
46 def assemble_dictionary(self, new_data_dict, old_data_dictionary={}):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
47 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
48 Create/update tools indices in the forward and backward dictionary
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
49 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
50 new_data_dict, reverse_dict = self.create_new_dict(new_data_dict)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
51 return new_data_dict, reverse_dict
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
52
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
53 def create_data_dictionary(self, words, old_data_dictionary={}):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
54 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
55 Create two dictionaries having tools names and their indexes
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
56 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
57 count = collections.Counter(words).most_common()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
58 dictionary = dict()
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
59 for index, (word, _) in enumerate(count):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
60 word = word.lstrip()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
61 word = word.rstrip()
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
62 dictionary[word] = len(dictionary) + 1
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
63 dictionary, reverse_dictionary = self.assemble_dictionary(dictionary, old_data_dictionary)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
64 return dictionary, reverse_dictionary
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
65
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
66 def decompose_paths(self, paths, dictionary):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
67 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
68 Decompose the paths to variable length sub-paths keeping the first tool fixed
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
69 """
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
70 max_len = 0
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
71 sub_paths_pos = list()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
72 for index, item in enumerate(paths):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
73 tools = item.split(",")
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
74 len_tools = len(tools)
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
75 if len_tools > max_len:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
76 max_len = len_tools
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
77 if len_tools < self.max_tool_sequence_len:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
78 sequence = tools[0: len_tools]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
79 tools_pos = [str(dictionary[str(tool_item)]) for tool_item in sequence]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
80 if len(tools_pos) > 1:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
81 sub_paths_pos.append(",".join(tools_pos))
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
82 sub_paths_pos = list(set(sub_paths_pos))
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
83 print("Max length of tools: ", max_len)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
84 return sub_paths_pos
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
85
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
86 def prepare_input_one_target_paths(self, dictionary, reverse_dictionary, paths):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
87 input_target_paths = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
88 compatible_tools = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
89 d_size = 0
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
90 for i, item in enumerate(paths):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
91 input_tools = item.split(",")
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
92 tool_seq = input_tools
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
93 i_tools = ",".join(tool_seq[0:-1])
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
94 last_i_tool = i_tools.split(",")[-1]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
95 if last_i_tool not in compatible_tools:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
96 compatible_tools[last_i_tool] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
97 t_tools = tool_seq[-1]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
98 if t_tools not in compatible_tools[last_i_tool]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
99 compatible_tools[last_i_tool].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
100 if i_tools not in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
101 input_target_paths[i_tools] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
102 if t_tools not in input_target_paths[i_tools]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
103 input_target_paths[i_tools].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
104 if i_tools not in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
105 input_target_paths[i_tools] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
106 if t_tools not in input_target_paths[i_tools]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
107 input_target_paths[i_tools].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
108 for item in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
109 d_size += len(input_target_paths[item])
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
110 print("Dataset size:", d_size)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
111 return input_target_paths, compatible_tools, d_size
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
112
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
113 def prepare_input_target_paths(self, dictionary, reverse_dictionary, paths):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
114 input_target_paths = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
115 compatible_tools = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
116 d_size = 0
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
117 for i, item in enumerate(paths):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
118 input_tools = item.split(",")
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
119 ctr = 0
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
120 for ctr in range(len(input_tools) - 1):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
121 # uncomment this for one token target idea
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
122 tool_seq = input_tools[0: ctr + 2]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
123 i_tools = ",".join(tool_seq[0:-1])
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
124 last_i_tool = i_tools.split(",")[-1]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
125 if last_i_tool not in compatible_tools:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
126 compatible_tools[last_i_tool] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
127 t_tools = tool_seq[-1]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
128 if t_tools not in compatible_tools[last_i_tool]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
129 compatible_tools[last_i_tool].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
130 if i_tools not in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
131 input_target_paths[i_tools] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
132 if t_tools not in input_target_paths[i_tools]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
133 input_target_paths[i_tools].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
134 if i_tools not in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
135 input_target_paths[i_tools] = list()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
136 if t_tools not in input_target_paths[i_tools]:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
137 input_target_paths[i_tools].append(t_tools)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
138 for item in input_target_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
139 d_size += len(input_target_paths[item])
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
140 print("Dataset size:", d_size)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
141 return input_target_paths, compatible_tools, d_size
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
142
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
143 def pad_paths_one_tool_target(self, multi_paths, compatible_tools, d_size, rev_dict, dictionary):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
144 d_size = len(multi_paths)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
145 input_mat = np.zeros([d_size, self.max_tool_sequence_len])
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
146 target_mat = np.zeros([d_size, len(dictionary) + 1])
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
147 train_counter = 0
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
148 for input_seq, target_seq_tools in list(multi_paths.items()):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
149 input_seq_tools = input_seq.split(",")
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
150 last_i_tool = input_seq_tools[-1]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
151 for id_pos, pos in enumerate(input_seq_tools):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
152 input_mat[train_counter][id_pos] = int(pos)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
153 if last_i_tool in compatible_tools:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
154 compatible_targets = compatible_tools[last_i_tool]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
155 for k, t_label in enumerate(target_seq_tools):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
156 target_mat[train_counter][int(t_label)] = 1
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
157 for c_tool in compatible_targets:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
158 target_mat[train_counter][int(c_tool)] = 1
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
159 train_counter += 1
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
160 print("Final data size: ", input_mat.shape, target_mat.shape)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
161 train_data, test_data, train_labels, test_labels = train_test_split(input_mat, target_mat, test_size=self.test_share, random_state=42)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
162 return train_data, train_labels, test_data, test_labels
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
163
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
164 def split_test_train_data(self, multilabels_paths):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
165 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
166 Split into test and train data randomly for each run
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
167 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
168 train_dict = dict()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
169 test_dict = dict()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
170 all_paths = multilabels_paths.keys()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
171 random.shuffle(list(all_paths))
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
172 split_number = int(self.test_share * len(all_paths))
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
173 for index, path in enumerate(list(all_paths)):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
174 if index < split_number:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
175 test_dict[path] = multilabels_paths[path]
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
176 else:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
177 train_dict[path] = multilabels_paths[path]
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
178 return train_dict, test_dict
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
179
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
180 def get_predicted_usage(self, data_dictionary, predicted_usage):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
181 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
182 Get predicted usage for tools
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
183 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
184 usage = dict()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
185 epsilon = 0.0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
186 # index 0 does not belong to any tool
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
187 usage[0] = epsilon
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
188 for k, v in data_dictionary.items():
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
189 try:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
190 usg = predicted_usage[k]
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
191 if usg < epsilon:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
192 usg = epsilon
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
193 usage[v] = usg
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
194 except Exception:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
195 usage[v] = epsilon
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
196 continue
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
197 return usage
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
198
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
199 def assign_class_weights(self, n_classes, predicted_usage):
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
200 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
201 Compute class weights using usage
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
202 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
203 class_weights = dict()
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
204 class_weights[str(0)] = 0.0
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
205 for key in range(1, n_classes + 1):
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
206 u_score = predicted_usage[key]
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
207 if u_score < 1.0:
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
208 u_score += 1.0
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
209 class_weights[key] = np.round(np.log(u_score), 6)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
210 return class_weights
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
211
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
212 def get_train_tool_labels_freq(self, train_paths, reverse_dictionary):
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
213 """
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
214 Get the frequency of last tool of each tool sequence
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
215 to estimate the frequency of tool sequences
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
216 """
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
217 last_tool_freq = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
218 freq_dict_names = dict()
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
219 for path in train_paths:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
220 tools_pos = np.where(path > 0)[0]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
221 path_pos = tools_pos
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
222 path_pos = [str(int(item)) for item in path_pos]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
223
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
224 for tool_pos in path_pos:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
225 if tool_pos not in last_tool_freq:
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
226 last_tool_freq[tool_pos] = 0
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
227 freq_dict_names[reverse_dictionary[int(tool_pos)]] = 0
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
228 last_tool_freq[tool_pos] += 1
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
229 freq_dict_names[reverse_dictionary[int(tool_pos)]] += 1
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
230 sorted_dict = dict(sorted(last_tool_freq.items(), key=lambda kv: kv[1], reverse=True))
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
231 return sorted_dict
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
232
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
233 def get_train_last_tool_freq(self, train_paths, reverse_dictionary):
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
234 """
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
235 Get the frequency of last tool of each tool sequence
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
236 to estimate the frequency of tool sequences
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
237 """
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
238 last_tool_freq = dict()
3
98bc44d17561 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents: 2
diff changeset
239 freq_dict_names = dict()
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
240 for path in train_paths:
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
241 tools_pos = np.where(path > 0)[0]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
242 path_pos = path[tools_pos]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
243 path_pos = [str(int(item)) for item in path_pos]
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
244 last_tool = path_pos[-1]
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
245 if last_tool not in last_tool_freq:
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
246 last_tool_freq[last_tool] = 0
3
98bc44d17561 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents: 2
diff changeset
247 freq_dict_names[reverse_dictionary[int(last_tool)]] = 0
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
248 last_tool_freq[last_tool] += 1
3
98bc44d17561 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents: 2
diff changeset
249 freq_dict_names[reverse_dictionary[int(last_tool)]] += 1
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
250 sorted_dict = dict(sorted(last_tool_freq.items(), key=lambda kv: kv[1], reverse=True))
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
251 return sorted_dict
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
252
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
253 def get_toolid_samples(self, train_data, l_tool_freq):
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
254 l_tool_tr_samples = dict()
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
255 for tool_id in l_tool_freq:
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
256 for index, tr_sample in enumerate(train_data):
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
257 last_tool_id = str(int(tr_sample[-1]))
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
258 if last_tool_id == tool_id:
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
259 if last_tool_id not in l_tool_tr_samples:
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
260 l_tool_tr_samples[last_tool_id] = list()
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
261 l_tool_tr_samples[last_tool_id].append(index)
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
262 return l_tool_tr_samples
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
263
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
264 def get_data_labels_matrices(self, workflow_paths, usage_df, cutoff_date, standard_connections, old_data_dictionary={}):
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
265 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
266 Convert the training and test paths into corresponding numpy matrices
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
267 """
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
268 processed_data, raw_paths = self.process_workflow_paths(workflow_paths)
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
269 dictionary, rev_dict = self.create_data_dictionary(processed_data, old_data_dictionary)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
270
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
271 num_classes = len(dictionary)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
272
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
273 print("Raw paths: %d" % len(raw_paths))
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
274 random.shuffle(raw_paths)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
275
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
276 print("Decomposing paths...")
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
277 all_unique_paths = self.decompose_paths(raw_paths, dictionary)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
278 random.shuffle(all_unique_paths)
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
279
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
280 print("Creating dictionaries...")
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
281 multilabels_paths, compatible_tools, d_size = self.prepare_input_target_paths(dictionary, rev_dict, all_unique_paths)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
282
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
283 print("Complete data: %d" % d_size)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
284
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
285 print("Padding train and test data...")
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
286 # pad training and test data with trailing zeros
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
287 train_data, train_labels, test_data, test_labels = self.pad_paths_one_tool_target(multilabels_paths, compatible_tools, d_size, rev_dict, dictionary)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
288
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
289 print("Train data: ", train_data.shape)
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
290 print("Test data: ", test_data.shape)
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
291
3
98bc44d17561 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents: 2
diff changeset
292 print("Estimating sample frequency...")
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
293 tr_tool_freq = self.get_train_tool_labels_freq(train_labels, rev_dict)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
294
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
295 # Predict tools usage
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
296 print("Predicting tools' usage...")
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
297 usage_pred = predict_tool_usage.ToolPopularity()
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
298 usage = usage_pred.extract_tool_usage(usage_df, cutoff_date, dictionary)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
299 tool_usage_prediction = usage_pred.get_pupularity_prediction(usage)
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
300 t_pred_usage = self.get_predicted_usage(dictionary, tool_usage_prediction)
0
22ebbac136c7 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
301 # get class weights using the predicted usage for each tool
2
50753817983a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents: 0
diff changeset
302 class_weights = self.assign_class_weights(num_classes, t_pred_usage)
5
9ec705bd11cb planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 24bab7a797f53fe4bcc668b18ee0326625486164
bgruening
parents: 4
diff changeset
303 return train_data, train_labels, test_data, test_labels, dictionary, rev_dict, class_weights, compatible_tools, tr_tool_freq