Newer
Older
blunck
committed
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
158
159
160
161
162
163
164
165
166
167
168
from feature_extraction import extract_features
import numpy as np
from sklearn import svm
from sklearn import tree
from sklearn import naive_bayes
from sklearn import linear_model
from sklearn.model_selection import cross_validate, GridSearchCV
from sklearn import metrics
import config
import csv
import utils
def run(corpus):
"""
Main part of the programm. Includes following steps:
- split the data
- extract features
- initiate classifiers
- train test, validate & tune classifiers
- evaluate results
"""
print("----------Configuration----------\nSelected Features: {}\n".format(config.feature_selection))
# split data set according to config
split_point = int(config.split_ratio*len(corpus))
train_set = corpus[:split_point]
test_set = corpus[split_point:]
# inputs (x)
train_inputs, test_inputs, features = extract_features(train_set, test_set)
# labels (y)
train_labels = np.array([int(el['LABEL']) for el in train_set]) # 1000 labels
test_labels = np.array([int(el['LABEL']) for el in test_set]) # 254 labels
# initiate classifiers
svm_clf = svm.SVC(C=0.1, gamma=0.001, kernel='linear') # best: C=0.1, gamma=0.001, kernel='linear'
tree_clf = tree.DecisionTreeClassifier(criterion='gini', max_depth=6, min_samples_leaf=2) # best: criterion='gini', max_depth=6, min_samples_leaf=2
nb_clf = naive_bayes.MultinomialNB(alpha=0.5) # best: alpha=0.5
lr_clf = linear_model.LogisticRegression(C=0.1) # best: C=0.1 #, 'penalty': 'l2'
# validation
if config.validate == True:
validate_multiple([svm_clf, tree_clf, nb_clf, lr_clf], train_inputs, train_labels, 10)
# tuning (takes ~24h!)
if config.tune == True:
tune_multiple(svm_clf, tree_clf, nb_clf, lr_clf, train_inputs, train_labels)
# training
if config.train or config.test or config.plot_weights or config.log_misclassifications:
train_multiple([svm_clf, tree_clf, nb_clf, lr_clf], train_inputs, train_labels)
# testing
if config.test == True:
test_multiple([svm_clf, tree_clf, nb_clf, lr_clf], test_inputs, test_labels)
# misclassification (SVM)
if config.log_misclassifications == True:
utils.get_misclassifications(svm_clf, test_set, test_inputs, test_labels)
# plot top features (SVM)
if config.plot_weights == True:
feature_names = utils.get_all_feature_names(features)
utils.plot_coefficients(svm_clf, feature_names)
def train_multiple(classifiers, train_inputs, train_labels):
for classifier in classifiers:
classifier.fit(train_inputs, train_labels)
def validate_multiple(classifiers, train_inputs, train_labels, folds):
"""
Runs cross validation for given classifiers.
Prints and writes results to file given in config.
"""
print("\n-------Cross Validation-------")
csv_file = open(config.validation_result_out, "a")
csv_writer = csv.DictWriter(csv_file, ["classifier", "features", "accuracy", "precision", "recall", "f1-score"])
for classifier in classifiers:
scoring = ['accuracy', 'precision', 'recall', 'f1']
scores = cross_validate(classifier, train_inputs, train_labels, scoring=scoring,
cv=folds, return_train_score=False)
acc = float("{0:.3f}".format(scores['test_accuracy'].mean()))
prec = float("{0:.3f}".format(scores['test_precision'].mean()))
recall = float("{0:.3f}".format(scores['test_recall'].mean()))
f1 = float("{0:.3f}".format(scores['test_f1'].mean()))
name = classifier.__str__().split("(")[0]
print("{}\nAccuracy: {}, Recall: {}, Precision: {}, F1-Score: {}\n".format(name, acc, recall, prec, f1))
csv_writer.writerow({"classifier":name, "features":config.feature_selection, "accuracy":acc,
"precision":prec, "recall":recall, "f1-score":f1
})
csv_file.close()
def get_best_params(classifier, param_grid, train_inputs, train_labels):
"""
Performs GridSearch to find best parameters
"""
print("{} \n".format(classifier))
grid_search = GridSearchCV(classifier, param_grid, cv=3)
grid_search.fit(train_inputs, train_labels)
print("Best parameters: {}".format(grid_search.best_params_))
print("Best score: {}".format(grid_search.best_score_))
def tune_multiple(svm, tree, nb, lr, train_inputs, train_labels):
svm_param_grid = {'C': [0.001, 0.01, 0.1, 1, 10],
'gamma' : [0.001, 0.01, 0.1, 1],
'kernel' : ['linear', 'rbf', 'poly']}
tree_param_grid = {'criterion' : ['gini', 'entropy'],
'max_depth': [9, 6, 3, None],
'max_features': [1, 2, 3, 4, 5, 6, 7, 8, 9, 500],
'min_samples_leaf': [1, 2, 3, 4, 5, 6, 7, 8, 9]}
nb_param_grid = {'alpha' : [0, 0.5, 1.0]}
lr_param_grid = {'penalty' : ['l1', 'l2'],
'C' : [0.001, 0.01, 0.1, 1, 10]}
# TODO: something better
get_best_params(svm, svm_param_grid, train_inputs, train_labels)
get_best_params(tree, tree_param_grid, train_inputs, train_labels)
get_best_params(nb, nb_param_grid, train_inputs, train_labels)
get_best_params(lr, lr_param_grid, train_inputs, train_labels)
def test_multiple(classifiers, test_inputs, test_labels):
"""
Evaluates given classifiers on test data.
Prints and writes results to file given in config.
"""
print("\n--------Test Data Scores----------")
csv_file = open(config.test_result_out, "a")
csv_writer = csv.DictWriter(csv_file, ["classifier", "features", "accuracy", "precision", "recall", "f1-score"])
for clf in classifiers:
acc = float("{0:.3f}".format(clf.score(test_inputs, test_labels)))
predictions = clf.predict(test_inputs)
f1 = float("{0:.3f}".format(metrics.f1_score(test_labels, predictions)))
recall = float("{0:.3f}".format(metrics.recall_score(test_labels, predictions)))
precision = float("{0:.3f}".format(metrics.precision_score(test_labels, predictions)))
name = clf.__str__().split("(")[0]
print("{}\nAccuracy: {}, Recall: {}, Precision: {}, F1-Score: {}\n".format(name, acc, recall, precision, f1))
csv_writer.writerow({"classifier":name, "features":config.feature_selection, "accuracy":acc,
"precision":precision, "recall":recall, "f1-score":f1
})
csv_file.close()