-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtf_idf.py
180 lines (145 loc) · 7.86 KB
/
tf_idf.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
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier
from lightgbm import LGBMClassifier
from xgboost import XGBClassifier
from sklearn.metrics import accuracy_score, recall_score, precision_score, roc_auc_score
from result_visualization import *
import numpy as np
import time
def run_tf_idf(train_text, train_label, test_text, test_label, models):
# TF-IDF
vectorizer = TfidfVectorizer(dtype=np.float64)
X_train = vectorizer.fit_transform(train_text)
X_test = vectorizer.transform(test_text)
metrics = {'model': [], 'accuracy': [], 'recall': [], 'precision': []}
if 'LogR' in models or 'all' in models:
# Logistic Regression model
start_time = time.time()
logR_model = LogisticRegression()
logR_model.fit(X_train, train_label)
logR_pred = logR_model.predict(X_test)
logR_pred_proba = logR_model.predict_proba(X_test)
end_time = time.time()
# metric
accuracy_log = accuracy_score(test_label, logR_pred)
recall_log = recall_score(test_label, logR_pred, average='macro')
precision_log = precision_score(test_label, logR_pred, average='macro')
auc_log = roc_auc_score(test_label, logR_pred_proba, multi_class='ovr')
metrics['model'].append('LogR')
metrics['accuracy'].append(accuracy_log)
metrics['recall'].append(recall_log)
metrics['precision'].append(precision_log)
print('Logistic Regression (TF-IDF) Accuracy: {:.3f}'.format(accuracy_log))
print('Recall: {:.3f}'.format(recall_log))
print('Precision: {:.3f}'.format(precision_log))
print('AUC: {:.3f}'.format(auc_log))
print('Running time: {:.3f}s'.format(end_time - start_time))
run_confusion_matrix(test_label, logR_pred,
label_name=['sadness', 'joy', 'love', 'anger', 'fear', 'surprise'],
title="Confusion Matrix of Logistic Regression based on TF-IDF",
save_path="./images/Confusion_Matrix_of_LogR_ON_TF_IDF.png")
if 'DT' in models or 'all' in models:
# Linear Regression model
start_time = time.time()
dt_model = DecisionTreeClassifier()
dt_model.fit(X_train, train_label)
dt_pred = dt_model.predict(X_test)
dt_pred_proba = dt_model.predict_proba(X_test)
end_time = time.time()
accuracy_dt = accuracy_score(test_label, dt_pred)
recall_dt = recall_score(test_label, dt_pred, average='macro')
precision_dt = precision_score(test_label, dt_pred, average='macro')
auc_dt = roc_auc_score(test_label, dt_pred_proba, multi_class='ovr')
metrics['model'].append('DT')
metrics['accuracy'].append(accuracy_dt)
metrics['recall'].append(recall_dt)
metrics['precision'].append(precision_dt)
print('Decision Tree (TF-IDF) Accuracy: {:.3f}'.format(accuracy_dt))
print('Recall: {:.3f}'.format(recall_dt))
print('Precision: {:.3f}'.format(precision_dt))
print('AUC: {:.3f}'.format(auc_dt))
print('Running time: {:.3f}s'.format(end_time - start_time))
run_confusion_matrix(test_label, dt_pred,
label_name=['sadness', 'joy', 'love', 'anger', 'fear', 'surprise'],
title="Confusion Matrix of Decision Tree based on TF-IDF",
save_path="./images/Confusion_Matrix_of_DT_ON_TF_IDF.png")
if 'XGB' in models or 'all' in models:
# XGB model
start_time = time.time()
xgb_model = XGBClassifier(probability=True)
xgb_model.fit(X_train, train_label)
xgb_pred = xgb_model.predict(X_test)
xgb_pred_proba = xgb_model.predict_proba(X_test)
end_time = time.time()
accuracy_xgb = accuracy_score(test_label, xgb_pred)
recall_xgb = recall_score(test_label, xgb_pred, average='macro')
precision_xgb = precision_score(test_label, xgb_pred, average='macro')
auc_xgb = roc_auc_score(test_label, xgb_pred_proba, multi_class='ovr')
metrics['model'].append('XGB')
metrics['accuracy'].append(accuracy_xgb)
metrics['recall'].append(recall_xgb)
metrics['precision'].append(precision_xgb)
print('XGB (TF-IDF) Accuracy: {:.3f}'.format(accuracy_xgb))
print('Recall: {:.3f}'.format(recall_xgb))
print('Precision: {:.3f}'.format(precision_xgb))
print('AUC: {:.3f}'.format(auc_xgb))
print('Running time: {:.3f}s'.format(end_time - start_time))
run_confusion_matrix(test_label, xgb_pred,
label_name=['sadness', 'joy', 'love', 'anger', 'fear', 'surprise'],
title="Confusion Matrix of XGB based on TF-IDF",
save_path="./images/Confusion_Matrix_of_XGB_ON_TF_IDF.png")
if 'LGB' in models or 'all' in models:
# LGBM model
start_time = time.time()
lgb_model = LGBMClassifier()
lgb_model.fit(X_train, train_label)
lgb_pred = lgb_model.predict(X_test)
lgb_pred_proba = lgb_model.predict_proba(X_test)
end_time = time.time()
accuracy_lgb = accuracy_score(test_label, lgb_pred)
recall_lgb = recall_score(test_label, lgb_pred, average='macro')
precision_lgb = precision_score(test_label, lgb_pred, average='macro')
auc_lgb = roc_auc_score(test_label, lgb_pred_proba, multi_class='ovr')
metrics['model'].append('LGB')
metrics['accuracy'].append(accuracy_lgb)
metrics['recall'].append(recall_lgb)
metrics['precision'].append(precision_lgb)
print('LGB (TF-IDF) Accuracy: {:.3f}'.format(accuracy_lgb))
print('Recall: {:.3f}'.format(recall_lgb))
print('Precision: {:.3f}'.format(precision_lgb))
print('AUC: {:.3f}'.format(auc_lgb))
print('Running time: {:.3f}s'.format(end_time - start_time))
run_confusion_matrix(test_label, lgb_pred,
label_name=['sadness', 'joy', 'love', 'anger', 'fear', 'surprise'],
title="Confusion Matrix of LGB based on TF-IDF",
save_path="./images/Confusion_Matrix_of_LGB_ON_TF_IDF.png")
if 'MNB' in models or 'all' in models:
# MultinomialNB model
start_time = time.time()
mnb_model = MultinomialNB()
mnb_model.fit(X_train, train_label)
mnb_pred = mnb_model.predict(X_test)
mnb_pred_proba = mnb_model.predict_proba(X_test)
end_time = time.time()
accuracy_mnb = accuracy_score(test_label, mnb_pred)
recall_mnb = recall_score(test_label, mnb_pred, average='macro')
precision_mnb = precision_score(test_label, mnb_pred, average='macro')
auc_mnb = roc_auc_score(test_label, mnb_pred_proba, multi_class='ovr')
metrics['model'].append('MNB')
metrics['accuracy'].append(accuracy_mnb)
metrics['recall'].append(recall_mnb)
metrics['precision'].append(precision_mnb)
print('MNB (TF-IDF) Accuracy: {:.3f}'.format(accuracy_mnb))
print('Recall: {:.3f}'.format(recall_mnb))
print('Precision: {:.3f}'.format(precision_mnb))
print('AUC: {:.3f}'.format(auc_mnb))
print('Running time: {:.3f}s'.format(end_time - start_time))
run_confusion_matrix(test_label, mnb_pred,
label_name=['sadness', 'joy', 'love', 'anger', 'fear', 'surprise'],
title="Confusion Matrix of MNB based on TF-IDF",
save_path="./images/Confusion_Matrix_of_MNB_ON_TF_IDF.png")
run_plot_metrics(metrics, save_path="./images/Comparison of Metrics in TF-IDF.png")