/
alternate_classifiers.py
157 lines (139 loc) · 5.6 KB
/
alternate_classifiers.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
from boto.s3.connection import S3Connection, Location
import datetime
import os
import time
import pickle
import diagnosis
from diagnosis.KeywordExtractor import *
import numpy as np
import re
import sklearn
import disease_label_table
from sklearn.pipeline import Pipeline
from sklearn.feature_extraction import DictVectorizer
from sklearn.multiclass import OneVsRestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import AdaBoostClassifier
from sklearn.svm import SVC
from diagnosis.utils import group_by, flatten
import warnings
import pymongo
from DataSet import fetch_datasets
def best_guess(classifier, X):
probs = classifier.predict_proba(X)[0]
p_max = max(probs)
result = {}
for i,p in enumerate(probs):
cutoff_ratio = 0.65
parents = disease_label_table.get_inferred_labels(classifier.classes_[i])
if p >= p_max * cutoff_ratio:
result[i] = max(p, result.get(i, 0))
for i2, label in enumerate(classifier.classes_):
if label in parents:
result[i2] = max(p, probs[i2], result.get(i2, 0))
return result.items()
def main():
print "Setting up"
classifiers = [
(OneVsRestClassifier(LogisticRegression(), n_jobs=-1), "OneVsRest(Logistic Regression)", True),
(DecisionTreeClassifier(), "Decision Tree Classifier", False),
(AdaBoostClassifier(DecisionTreeClassifier()), "AdaBoost(Decision Tree Classifier)", False),
(OneVsRestClassifier(SVC(probability=True), n_jobs=-1), "OneVsRest(SVC)", True)
]
with open('ontologies.p') as f:
keywords = pickle.load(f)
categories = set([
'hm/disease',
'biocaster/pathogens',
'biocaster/diseases',
'biocaster/symptoms',
'symp/symptoms',
'eha/symptom',
'eha/mode of transmission',
'eha/environmental factors',
'eha/vector',
'eha/occupation',
'eha/control measures',
'eha/description of infected',
'eha/disease category',
'eha/host',
'eha/host use',
'eha/symptom',
'eha/disease',
'eha/location',
'eha/transmission',
'eha/zoonotic type',
'eha/risk',
'wordnet/season',
'wordnet/climate',
'wordnet/pathogens',
'wordnet/hosts',
'wordnet/mod/severe',
'wordnet/mod/painful',
'wordnet/mod/large',
'wordnet/mod/rare',
'doid/has_symptom',
'doid/symptoms',
'doid/transmitted_by',
'doid/located_in',
'doid/diseases',
'doid/results_in',
'doid/has_material_basis_in',
'usgs/terrain'
])
keyword_array = [
keyword_obj for keyword_obj in keywords
if keyword_obj['category'] in categories
]
feature_extractor = Pipeline([
('kwext', KeywordExtractor(keyword_array)),
('link', LinkedKeywordAdder(keyword_array)),
('limit', LimitCounts(1)),
])
print "Fetching datasets"
time_offset_test_set, mixed_test_set, training_set = fetch_datasets()
print "Setting up vectorizers and extractors"
time_offset_test_set.feature_extractor =mixed_test_set.feature_extractor =training_set.feature_extractor = feature_extractor
my_dict_vectorizer = DictVectorizer(sparse=False).fit(training_set.get_feature_dicts())
time_offset_test_set.dict_vectorizer = mixed_test_set.dict_vectorizer = training_set.dict_vectorizer = my_dict_vectorizer
print "Removing zero feature vectors"
time_offset_test_set.remove_zero_feature_vectors()
mixed_test_set.remove_zero_feature_vectors()
training_set.remove_zero_feature_vectors()
feature_array = np.array(training_set.get_feature_vectors())
label_array = np.array(training_set.get_labels())
with warnings.catch_warnings():
# The updated version of scikit will spam warnings here.
warnings.simplefilter("ignore")
for (my_classifier, classifier_label, add_parents) in classifiers:
print("Fitting classifier: " + classifier_label)
before_time = time.clock()
my_classifier.fit(feature_array, label_array)
after_time = time.clock()
print("Training time: " + str(after_time - before_time))
print("Testing:")
before_time = time.clock()
for data_set, ds_label, print_label_breakdown in [
(training_set, "Training set", False),
(time_offset_test_set, "Time offset set", True),
(mixed_test_set, "Mixed test set", False),
]:
y_true = data_set.get_labels()
if add_parents:
guesses = [best_guess(my_classifier, vector) for vector in data_set.get_feature_vectors()]
y_pred = []
for guess in guesses:
y_pred.append([my_classifier.classes_[i] for (i, p) in guess])
else:
y_pred = my_classifier.predict(data_set.get_feature_vectors())
print (ds_label + " (macro) \n"
"precision: %s recall: %s f-score: %s") %\
sklearn.metrics.precision_recall_fscore_support(y_true, y_pred, average='macro')[0:3]
print (ds_label + " (micro) \n"
"precision: %s recall: %s f-score: %s") %\
sklearn.metrics.precision_recall_fscore_support(y_true, y_pred, average='micro')[0:3]
after_time = time.clock()
print("Testing time: " + str(after_time - before_time))
if __name__ == "__main__":
main()