Mercurial > repos > bgruening > create_tool_recommendation_model
annotate utils.py @ 4:afec8c595124 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
author | bgruening |
---|---|
date | Tue, 07 Jul 2020 03:25:49 -0400 |
parents | 5b3c08710e47 |
children | 4f7e6612906b |
rev | line source |
---|---|
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
1 import numpy as np |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
2 import json |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
3 import h5py |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
4 import random |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
5 from numpy.random import choice |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
6 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
7 from keras import backend as K |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
8 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
9 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
10 def read_file(file_path): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
11 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
12 Read a file |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
13 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
14 with open(file_path, "r") as json_file: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
15 file_content = json.loads(json_file.read()) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
16 return file_content |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
17 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
18 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
19 def format_tool_id(tool_link): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
20 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
21 Extract tool id from tool link |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
22 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
23 tool_id_split = tool_link.split("/") |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
24 tool_id = tool_id_split[-2] if len(tool_id_split) > 1 else tool_link |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
25 return tool_id |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
26 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
27 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
28 def set_trained_model(dump_file, model_values): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
29 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
30 Create an h5 file with the trained weights and associated dicts |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
31 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
32 hf_file = h5py.File(dump_file, 'w') |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
33 for key in model_values: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
34 value = model_values[key] |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
35 if key == 'model_weights': |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
36 for idx, item in enumerate(value): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
37 w_key = "weight_" + str(idx) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
38 if w_key in hf_file: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
39 hf_file.modify(w_key, item) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
40 else: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
41 hf_file.create_dataset(w_key, data=item) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
42 else: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
43 if key in hf_file: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
44 hf_file.modify(key, json.dumps(value)) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
45 else: |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
46 hf_file.create_dataset(key, data=json.dumps(value)) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
47 hf_file.close() |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
48 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
49 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
50 def weighted_loss(class_weights): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
51 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
52 Create a weighted loss function. Penalise the misclassification |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
53 of classes more with the higher usage |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
54 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
55 weight_values = list(class_weights.values()) |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
56 weight_values.extend(weight_values) |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
57 def weighted_binary_crossentropy(y_true, y_pred): |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
58 # add another dimension to compute dot product |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
59 expanded_weights = K.expand_dims(weight_values, axis=-1) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
60 return K.dot(K.binary_crossentropy(y_true, y_pred), expanded_weights) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
61 return weighted_binary_crossentropy |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
62 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
63 |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
64 def balanced_sample_generator(train_data, train_labels, batch_size, l_tool_tr_samples, reverse_dictionary): |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
65 while True: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
66 dimension = train_data.shape[1] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
67 n_classes = train_labels.shape[1] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
68 tool_ids = list(l_tool_tr_samples.keys()) |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
69 random.shuffle(tool_ids) |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
70 generator_batch_data = np.zeros([batch_size, dimension]) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
71 generator_batch_labels = np.zeros([batch_size, n_classes]) |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
72 generated_tool_ids = choice(tool_ids, batch_size) |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
73 for i in range(batch_size): |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
74 random_toolid = generated_tool_ids[i] |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
75 sample_indices = l_tool_tr_samples[str(random_toolid)] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
76 random_index = random.sample(range(0, len(sample_indices)), 1)[0] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
77 random_tr_index = sample_indices[random_index] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
78 generator_batch_data[i] = train_data[random_tr_index] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
79 generator_batch_labels[i] = train_labels[random_tr_index] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
80 yield generator_batch_data, generator_batch_labels |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
81 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
82 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
83 def compute_precision(model, x, y, reverse_data_dictionary, usage_scores, actual_classes_pos, topk, standard_conn, last_tool_id, lowest_tool_ids): |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
84 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
85 Compute absolute and compatible precision |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
86 """ |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
87 pred_t_name = "" |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
88 top_precision = 0.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
89 mean_usage = 0.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
90 usage_wt_score = list() |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
91 pub_precision = 0.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
92 lowest_pub_prec = 0.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
93 lowest_norm_prec = 0.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
94 pub_tools = list() |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
95 actual_next_tool_names = list() |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
96 test_sample = np.reshape(x, (1, len(x))) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
97 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
98 # predict next tools for a test path |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
99 prediction = model.predict(test_sample, verbose=0) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
100 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
101 # divide the predicted vector into two halves - one for published and |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
102 # another for normal workflows |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
103 nw_dimension = prediction.shape[1] |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
104 half_len = int(nw_dimension / 2) |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
105 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
106 # predict tools |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
107 prediction = np.reshape(prediction, (nw_dimension,)) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
108 # get predictions of tools from published workflows |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
109 standard_pred = prediction[:half_len] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
110 # get predictions of tools from normal workflows |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
111 normal_pred = prediction[half_len:] |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
112 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
113 standard_prediction_pos = np.argsort(standard_pred, axis=-1) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
114 standard_topk_prediction_pos = standard_prediction_pos[-topk] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
115 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
116 normal_prediction_pos = np.argsort(normal_pred, axis=-1) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
117 normal_topk_prediction_pos = normal_prediction_pos[-topk] |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
118 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
119 # get true tools names |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
120 for a_t_pos in actual_classes_pos: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
121 if a_t_pos > half_len: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
122 t_name = reverse_data_dictionary[int(a_t_pos - half_len)] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
123 else: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
124 t_name = reverse_data_dictionary[int(a_t_pos)] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
125 actual_next_tool_names.append(t_name) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
126 last_tool_name = reverse_data_dictionary[x[-1]] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
127 # compute scores for published recommendations |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
128 if standard_topk_prediction_pos in reverse_data_dictionary: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
129 pred_t_name = reverse_data_dictionary[int(standard_topk_prediction_pos)] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
130 if last_tool_name in standard_conn: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
131 pub_tools = standard_conn[last_tool_name] |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
132 if pred_t_name in pub_tools: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
133 pub_precision = 1.0 |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
134 # count precision only when there is actually true published tools |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
135 if last_tool_id in lowest_tool_ids: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
136 lowest_pub_prec = 1.0 |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
137 else: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
138 lowest_pub_prec = np.nan |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
139 if standard_topk_prediction_pos in usage_scores: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
140 usage_wt_score.append(np.log(usage_scores[standard_topk_prediction_pos] + 1.0)) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
141 else: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
142 # count precision only when there is actually true published tools |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
143 # else set to np.nan. Set to 0 only when there is wrong prediction |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
144 pub_precision = np.nan |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
145 lowest_pub_prec = np.nan |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
146 # compute scores for normal recommendations |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
147 if normal_topk_prediction_pos in reverse_data_dictionary: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
148 pred_t_name = reverse_data_dictionary[int(normal_topk_prediction_pos)] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
149 if pred_t_name in actual_next_tool_names: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
150 if normal_topk_prediction_pos in usage_scores: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
151 usage_wt_score.append(np.log(usage_scores[normal_topk_prediction_pos] + 1.0)) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
152 top_precision = 1.0 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
153 if last_tool_id in lowest_tool_ids: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
154 lowest_norm_prec = 1.0 |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
155 else: |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
156 lowest_norm_prec = np.nan |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
157 if len(usage_wt_score) > 0: |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
158 mean_usage = np.mean(usage_wt_score) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
159 return mean_usage, top_precision, pub_precision, lowest_pub_prec, lowest_norm_prec |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
160 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
161 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
162 def get_lowest_tools(l_tool_freq, fraction=0.25): |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
163 l_tool_freq = dict(sorted(l_tool_freq.items(), key=lambda kv: kv[1], reverse=True)) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
164 tool_ids = list(l_tool_freq.keys()) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
165 lowest_ids = tool_ids[-int(len(tool_ids) * fraction):] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
166 return lowest_ids |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
167 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
168 |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
169 def verify_model(model, x, y, reverse_data_dictionary, usage_scores, standard_conn, lowest_tool_ids, topk_list=[1, 2, 3]): |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
170 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
171 Verify the model on test data |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
172 """ |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
173 print("Evaluating performance on test data...") |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
174 print("Test data size: %d" % len(y)) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
175 size = y.shape[0] |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
176 precision = np.zeros([len(y), len(topk_list)]) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
177 usage_weights = np.zeros([len(y), len(topk_list)]) |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
178 epo_pub_prec = np.zeros([len(y), len(topk_list)]) |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
179 epo_lowest_tools_pub_prec = list() |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
180 epo_lowest_tools_norm_prec = list() |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
181 lowest_counter = 0 |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
182 # loop over all the test samples and find prediction precision |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
183 for i in range(size): |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
184 lowest_pub_topk = list() |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
185 lowest_norm_topk = list() |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
186 actual_classes_pos = np.where(y[i] > 0)[0] |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
187 test_sample = x[i, :] |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
188 last_tool_id = str(int(test_sample[-1])) |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
189 for index, abs_topk in enumerate(topk_list): |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
190 usg_wt_score, absolute_precision, pub_prec, lowest_p_prec, lowest_n_prec = compute_precision(model, test_sample, y, reverse_data_dictionary, usage_scores, actual_classes_pos, abs_topk, standard_conn, last_tool_id, lowest_tool_ids) |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
191 precision[i][index] = absolute_precision |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
192 usage_weights[i][index] = usg_wt_score |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
193 epo_pub_prec[i][index] = pub_prec |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
194 lowest_pub_topk.append(lowest_p_prec) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
195 lowest_norm_topk.append(lowest_n_prec) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
196 epo_lowest_tools_pub_prec.append(lowest_pub_topk) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
197 epo_lowest_tools_norm_prec.append(lowest_norm_topk) |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
198 if last_tool_id in lowest_tool_ids: |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
199 lowest_counter += 1 |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
200 mean_precision = np.mean(precision, axis=0) |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
201 mean_usage = np.mean(usage_weights, axis=0) |
4
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
202 mean_pub_prec = np.nanmean(epo_pub_prec, axis=0) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
203 mean_lowest_pub_prec = np.nanmean(epo_lowest_tools_pub_prec, axis=0) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
204 mean_lowest_norm_prec = np.nanmean(epo_lowest_tools_norm_prec, axis=0) |
afec8c595124
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 65d36f271296a38deeceb0d0e8d471b2898ee8f4"
bgruening
parents:
3
diff
changeset
|
205 return mean_usage, mean_precision, mean_pub_prec, mean_lowest_pub_prec, mean_lowest_norm_prec, lowest_counter |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
206 |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
207 |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
208 def save_model(results, data_dictionary, compatible_next_tools, trained_model_path, class_weights, standard_connections): |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
209 # save files |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
210 trained_model = results["model"] |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
211 best_model_parameters = results["best_parameters"] |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
212 model_config = trained_model.to_json() |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
213 model_weights = trained_model.get_weights() |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
214 model_values = { |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
215 'data_dictionary': data_dictionary, |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
216 'model_config': model_config, |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
217 'best_parameters': best_model_parameters, |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
218 'model_weights': model_weights, |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
219 "compatible_tools": compatible_next_tools, |
3
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
220 "class_weights": class_weights, |
5b3c08710e47
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit c635df659fe1835679438589ded43136b0e515c6"
bgruening
parents:
2
diff
changeset
|
221 "standard_connections": standard_connections |
0
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
222 } |
9bf25dbe00ad
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff
changeset
|
223 set_trained_model(trained_model_path, model_values) |