-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathNews Headlines Data Product.py
359 lines (312 loc) · 13.3 KB
/
News Headlines Data Product.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
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report,confusion_matrix
from sklearn.model_selection import KFold
import numpy as np
import xgboost as xg
from sklearn.metrics import mean_squared_error as MSE
import pandas as pd
from sklearn.feature_extraction import text
from sklearn.feature_extraction.text import TfidfVectorizer
from nltk.tokenize import RegexpTokenizer
from nltk.stem.snowball import SnowballStemmer
import chardet
from nltk.stem import WordNetLemmatizer
from joblib import dump, load
from numpy import savetxt
from pandas import DataFrame
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import AdaBoostClassifier
from sklearn.ensemble import BaggingClassifier
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.experimental import enable_hist_gradient_boosting
from sklearn.ensemble import HistGradientBoostingClassifier
from sklearn.svm import SVC
from sklearn.ensemble import VotingClassifier
from sklearn.decomposition import TruncatedSVD
from sklearn.decomposition import LatentDirichletAllocation
from sklearn.neighbors import KNeighborsClassifier
from sklearn.cluster import KMeans
from sklearn.cluster import DBSCAN
from sklearn.manifold import TSNE
from sklearn.metrics import accuracy_score
from sklearn.model_selection import cross_val_score
# Read CSV
Raw_headlines = pd.read_csv('abcnews-date-text.csv')
print(Raw_headlines.info())
Process_Headline = Raw_headlines.copy(deep = True)
#drop duplicated head_lines.
Process_Headline.drop_duplicates(subset ="headline_text", keep = False, inplace = True)
print(Process_Headline.info())
# Determine tokenizer
tokenizer = RegexpTokenizer(r'[a-zA-Z\']+')
# Determine Stemming method
stemmer = SnowballStemmer('english')
lemmas = WordNetLemmatizer()
#get the stop word list
# explicly add punciation to the stop word list from the SKLearn
punc = ['.', ',', '"', "'", '?', '!', ':', ';', '(', ')', '[', ']', '{', '}',"%"]
StopWord = text.ENGLISH_STOP_WORDS.union(punc)
# Process the headline to lower letter
Process_Headline['headline_text'] = Process_Headline['headline_text'].str.lower()
#Lemmatizer approach
Process_Headline['headline_text'] = Process_Headline['headline_text'].apply(lambda x: ' '.join([lemmas.lemmatize(word) for word in tokenizer.tokenize(x)]))
#Stemmar approach
#Process_Headline['headline_text'] = Process_Headline['headline_text'].apply(lambda x: ' '.join([stemmer.stem(word) for word in tokenizer.tokenize(x)]))
headline_content = Process_Headline['headline_text'].values
vectorizer = TfidfVectorizer(stop_words = StopWord, max_features = 40000)
TD_IDF_DATA = vectorizer.fit_transform(headline_content)
TD_IDF_DATA = TD_IDF_DATA.sorted_indices()
#Failed AE
'''
### Encoder
encoder = Sequential()
encoder.add(Dense(9000,input_dim=len(word_features2),activation="relu"))
encoder.add(Dense(8000,activation="relu"))
encoder.add(Dense(7000,activation="relu"))
### Decoder
decoder = Sequential()
decoder.add(Dense(8000,input_shape=[7000],activation='relu'))
decoder.add(Dense(9000,activation='relu'))
decoder.add(Dense(len(word_features2),activation ="sigmoid"))
### Autoencoder
autoencoder = Sequential([encoder,decoder])
autoencoder.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])
autoencoder.fit(Stem_header,Stem_header,epochs=200,batch_size=128,verbose = 1)
AE_Headlines = encoder.predict(Stem_header)
print(AE_Headlines[:10])
'''
#Dimensionally reduce all the featres into 100 topics.
'''
LDA_Model = load('LDA_Model.joblib')
LDA_Data = pd.read_csv('LDA_Data.csv')
LDA_Data = LDA_Data.sample(n = 500000)
LDA_Data.sort_index()
'''
print("Construct LDA model")
LDA_Model=LatentDirichletAllocation(n_components=10,learning_method='online',random_state=42,max_iter=3,verbose= 1)
print("Transforming orginal data to LDA data")
LDA_Data=LDA_Model.fit_transform(TD_IDF_DATA)
print("Saving LDA Model")
dump(LDA_Model, 'LDA_Model.joblib')
savetxt('LDA_Data.csv', LDA_Data, delimiter=',')
print("Saved LDA Model")
'''
LDA_Model = load('LDA_Model.joblib')
LDA_Data = pd.read_csv('LDA_Data.csv', header = None)
'''
print('tsne2')
TSNE_Model = TSNE(n_components=2, n_jobs = 4, verbose = 2, random_state = 42).fit_transform(LDA_Data)
dump(TSNE_Model, 'TSNE_Model.joblib')
savetxt('TSNE_Model.csv', TSNE_Model, delimiter=',')
print(TSNE_Model[:10])
'''
print('tsne3')
TSNE_Model_3 = TSNE(n_components=3, n_jobs = 4, verbose = 2, random_state = 42).fit_transform(LDA_Data)
dump(TSNE_Model_3, 'TSNE_Model_3.joblib')
savetxt('TSNE_Model_3.csv', TSNE_Model_3, delimiter=',')
print(TSNE_Model_3[:10])
'''
vocab = vectorizer.get_feature_names()
for i, comp in enumerate(LDA_Model.components_):
vocab_comp = zip(vocab, comp)
sorted_words = sorted(vocab_comp, key= lambda x:x[1], reverse=True)[:10]
print("Topic "+str(i)+": ")
for t in sorted_words:
print(t[0]+', ', end = '')
print("\n")
'''
LDA_Data_5K = LDA_Data.sample(n = 5000)
LDA_Data_10K = LDA_Data.sample(n = 10000)
LDA_Data_20K = LDA_Data.sample(n = 20000)
LDA_Data_50K = LDA_Data.sample(n = 50000)
LDA_Data_100K = LDA_Data.sample(n = 100000)
LDA_Data_150K = LDA_Data.sample(n = 150000)
LDA_Data_250K = LDA_Data.sample(n = 250000)
LDA_Data_500K = LDA_Data.sample(n = 500000)
'''
print("Construct KMeans Model")
KMeans_Clustering = KMeans(n_clusters=10, random_state=42, verbose = 2, n_jobs = 3)
print("Clustering KMeans Model")
KMeans_label = KMeans_Clustering.fit_predict(TSNE_Model)
savetxt('KMeans_label.csv', KMeans_label,delimiter=',')
dump(KMeans_Clustering, 'KMeans_Clustering.joblib')
'''
KMeans_label_5k = KMeans_Clustering.fit_predict(LDA_Data_5K)
KMeans_label_10k = KMeans_Clustering.fit_predict(LDA_Data_10K)
KMeans_label_20k = KMeans_Clustering.fit_predict(LDA_Data_20K)
KMeans_label_50k = KMeans_Clustering.fit_predict(LDA_Data_50K)
KMeans_label_100k = KMeans_Clustering.fit_predict(LDA_Data_100K)
KMeans_label_150k = KMeans_Clustering.fit_predict(LDA_Data_150K)
KMeans_label_250k = KMeans_Clustering.fit_predict(LDA_Data_250K)
KMeans_label_500k = KMeans_Clustering.fit_predict(LDA_Data_500K)
print(KMeans_label[:1000])
print(KMeans_Clustering.labels_.tolist().count(0))
print(KMeans_Clustering.labels_.tolist().count(1))
print(KMeans_Clustering.labels_.tolist().count(2))
print(KMeans_Clustering.labels_.tolist().count(3))
print(KMeans_Clustering.labels_.tolist().count(4))
print(KMeans_Clustering.labels_.tolist().count(5))
print(KMeans_Clustering.labels_.tolist().count(6))
print(KMeans_Clustering.labels_.tolist().count(7))
print(KMeans_Clustering.labels_.tolist().count(8))
print(KMeans_Clustering.labels_.tolist().count(9))
print(KMeans_Clustering.cluster_centers_)
exit()
print("Construct DBSCAN Model")
DBSCAN_Clustering = DBSCAN()
print("Clustering DBSCAN Model")
DBSCAN_Label = DBSCAN_Clustering.fit_predict(LDA_Data)
print("Saving DBSCAN_Clustering Model")
dump(DBSCAN_Clustering, 'DBSCAN_Clustering.joblib')
savetxt('DBSCAN_Label.csv', DBSCAN_Label, delimiter=',')
print("Saved DBSCAN_Clustering Model")
print(DBSCAN_Label[:1000])
print("2015")
print(KMeans_label[:10])
KMeans_label_2015 = KMeans_label[925481:1002770+1]
KMeans_label_2015 = KMeans_label_2015.iloc[:,0]
print(KMeans_label_2015.tolist().count(0))
print(KMeans_label_2015.tolist().count(1))
print(KMeans_label_2015.tolist().count(2))
print(KMeans_label_2015.tolist().count(3))
print(KMeans_label_2015.tolist().count(4))
print(KMeans_label_2015.tolist().count(5))
print(KMeans_label_2015.tolist().count(6))
print(KMeans_label_2015.tolist().count(7))
print(KMeans_label_2015.tolist().count(8))
print(KMeans_label_2015.tolist().count(9))
print("2016")
KMeans_label_2016 = KMeans_label[1002771:1064344+1]
KMeans_label_2016 = KMeans_label_2016.iloc[:,0]
print(KMeans_label_2016.tolist().count(0))
print(KMeans_label_2016.tolist().count(1))
print(KMeans_label_2016.tolist().count(2))
print(KMeans_label_2016.tolist().count(3))
print(KMeans_label_2016.tolist().count(4))
print(KMeans_label_2016.tolist().count(5))
print(KMeans_label_2016.tolist().count(6))
print(KMeans_label_2016.tolist().count(7))
print(KMeans_label_2016.tolist().count(8))
print(KMeans_label_2016.tolist().count(9))
print("2017")
KMeans_label_2017 = KMeans_label[1064345:1111855+1]
KMeans_label_2017 = KMeans_label_2017.iloc[:,0]
print(KMeans_label_2017.tolist().count(0))
print(KMeans_label_2017.tolist().count(1))
print(KMeans_label_2017.tolist().count(2))
print(KMeans_label_2017.tolist().count(3))
print(KMeans_label_2017.tolist().count(4))
print(KMeans_label_2017.tolist().count(5))
print(KMeans_label_2017.tolist().count(6))
print(KMeans_label_2017.tolist().count(7))
print(KMeans_label_2017.tolist().count(8))
print(KMeans_label_2017.tolist().count(9))
print("2018")
KMeans_label_2018 = KMeans_label[1111856:1151957+1]
KMeans_label_2018 = KMeans_label_2018.iloc[:,0]
print(KMeans_label_2018.tolist().count(0))
print(KMeans_label_2018.tolist().count(1))
print(KMeans_label_2018.tolist().count(2))
print(KMeans_label_2018.tolist().count(3))
print(KMeans_label_2018.tolist().count(4))
print(KMeans_label_2018.tolist().count(5))
print(KMeans_label_2018.tolist().count(6))
print(KMeans_label_2018.tolist().count(7))
print(KMeans_label_2018.tolist().count(8))
print(KMeans_label_2018.tolist().count(9))
exit()
'''
train_X, test_X, train_y, test_y = train_test_split(TSNE_Model, KMeans_label, test_size=0.15, shuffle=True, random_state = 0)
print("Start KNN")
KNN_Model = KNeighborsClassifier(n_neighbors=7, n_jobs = 3)
print("Fit KNN")
KNN_Model.fit(train_X, train_y)
dump(KNN_Model, 'KNN_Model.joblib')
print("Predict KNN")
KNN_pred = KNN_Model.predict(test_X)
savetxt('KNN_pred.csv', KNN_pred, delimiter=',')
print(classification_report(test_y, KNN_pred))
print(confusion_matrix(test_y, KNN_pred))
RF_Model = RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini', max_features='auto', oob_score=True, random_state=42, verbose=2, warm_start=False, n_jobs = 4)
print("Fit RF_Model")
RF_Model.fit(train_X, train_y)
dump(RF_Model, 'RF_Model.joblib')
print("Predict RF_Model")
RF_pred = RF_Model.predict(test_X)
savetxt('RF_pred.csv', RF_pred, delimiter=',')
print(classification_report(test_y, RF_pred))
print(confusion_matrix(test_y, RF_pred))
#Cross validation of XGB
'''
kf = KFold(n_splits=20)
XGB_Model = xg.XGBClassifier( verbosity = 1, random_state = 42, n_jobs = 4)
print("Fit XGB_Model")
results = cross_val_score(XGB_Model, TSNE_Model, KMeans_label, cv=kf)
print("Accuracy: %.2f%% (%.2f%%)" % (results.mean()*100, results.std()*100))
exit()
'''
XGB_Model = xg.XGBClassifier( verbosity = 1, random_state = 42, n_jobs = 4)
print("Fit XGB_Model")
XGB_Model.fit(train_X, train_y)
dump(XGB_Model, 'XGB_Model.joblib')
print("Predict XGB_Model")
XGB_pred = XGB_Model.predict(test_X)
savetxt('XGB_pred.csv', XGB_pred, delimiter=',')
print(classification_report(test_y, XGB_pred))
print(confusion_matrix(test_y, XGB_pred))
XGBRF_Model = xg.XGBRFClassifier( verbosity =2, random_state = 42, n_jobs = 3)
print("Fit XGBRF_Model")
XGBRF_Model.fit(train_X, train_y)
dump(XGBRF_Model, 'XGBRF_Model.joblib')
print("Predict XGBRF_Model")
print("Predict XGBRF_Model")
XGBRF_pred = XGBRF_Model.predict(test_X)
savetxt('XGBRF_pred.csv', XGBRF_pred, delimiter=',')
print(classification_report(test_y, XGBRF_pred))
print(confusion_matrix(test_y, XGBRF_pred))
print("Create ADA_Model")
ADA_Model = AdaBoostClassifier(random_state = 42, n_estimators = 100)
print("Fit ADA_Model")
ADA_Model.fit(train_X, train_y)
dump(ADA_Model, 'ADA_Model.joblib')
ADA_pred = ADA_Model.predict(test_X)
savetxt('ADA_pred.csv', ADA_pred, delimiter=',')
print(classification_report(test_y, ADA_pred))
print(confusion_matrix(test_y, ADA_pred))
print("Create BAG_Model")
BAG_Model = BaggingClassifier(base_estimator=SVC(), n_jobs = 4, random_state = 42, verbose = 2)
print("Fit BAG_Model")
BAG_Model.fit(train_X, train_y)
dump(BAG_Model, 'BAG_Model.joblib')
BAG_pred = BAG_Model.predict(test_X)
savetxt('BAG_pred.csv', BAG_pred, delimiter=',')
print(classification_report(test_y, BAG_pred))
print(confusion_matrix(test_y, BAG_pred))
print("Create GBC_Model")
GBC_Model = GradientBoostingClassifier(random_state = 42, verbose = 2)
print("Fit GBC_Model")
GBC_Model.fit(train_X, train_y)
dump(GBC_Model, 'GBC_Model.joblib')
GBC_pred = GBC_Model.predict(test_X)
savetxt('GBC_pred.csv', GBC_pred, delimiter=',')
print(classification_report(test_y, GBC_pred))
print(confusion_matrix(test_y, GBC_pred))
print("Create HGBC_Model")
HGBC_Model = HistGradientBoostingClassifier(random_state = 42, verbose = 2)
print("Fit HGBC_Model")
HGBC_Model.fit(train_X, train_y)
dump(HGBC_Model, 'HGBC_Model.joblib')
HGBC_pred = HGBC_Model.predict(test_X)
savetxt('HGBC_pred.csv', HGBC_pred, delimiter=',')
print(classification_report(test_y, HGBC_pred))
print(confusion_matrix(test_y, HGBC_pred))
print("Create SVC_Model")
SVC_Model = SVC(random_state = 42, verbose = 2)
print("Fit SVC_Model")
SVC_Model.fit(train_X, train_y)
dump(SVC_Model, 'SVC_Model.joblib')
SVC_pred = SVC_Model.predict(test_X)
savetxt('SVC_pred.csv', SVC_pred, delimiter=',')
print(classification_report(test_y, SVC_pred))
print(confusion_matrix(test_y, SVC_pred))