-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
151 lines (101 loc) · 4.29 KB
/
train.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
from ServerModel.model import MultiClassLogisticRegression
import numpy as np
import pandas as pd
import re
import pickle as pk
from sklearn.feature_extraction.text import TfidfVectorizer, TfidfTransformer
from sklearn.preprocessing import LabelEncoder
class Server_hidden:
# evaluate the data and update the weights if it gives better metrics
def test(self,tfidf,encoder,log_reg,log_reg1):
data = pd.read_csv("test_data.csv")
data['TEXT'] = [_normalize_text(s) for s in data['TITLE']]
test_data_x= tfidf.transform(data['TEXT'])
test_data_x=test_data_x.toarray()
encoder = LabelEncoder()
y = encoder.transform(data['CATEGORY'])
rr=log_reg.predict_classes(test_data_x)
rr1=log_reg1.predict_classes(test_data_x)
print("Accuracy score: ", accuracy_score(y, rr))
print("Recall score: ", recall_score(y, rr, average = 'weighted'))
print("Precision score: ", precision_score(y, rr, average = 'weighted'))
print("F1 score: ", f1_score(y, rr, average = 'weighted'))
old=0
new=0
if(accuracy_score(y, rr)>accuracy_score(y, rr1)):
new+=1
else:
old+=1
if(recall_score(y, rr, average = 'weighted')>recall_score(y, rr1, average = 'weighted')):
new+=1
else:
old+=1
if(precision_score(y, rr, average = 'weighted')>precision_score(y, rr1, average = 'weighted')):
new+=1
else:
old+=1
if(f1_score(y, rr, average = 'weighted')>f1_score(y, rr1, average = 'weighted')):
new+=1
else:
old+=1
if(new>old):
new_weights = np.genfromtxt(new_weights_file ,delimiter=",",dtype=float)
np.savetxt("ServerModel/weights.csv", new_weights, delimiter=",")
print("Weights Updated!")
else:
print("Weights Did not Change")
return
def train(self, N):
print('Begin Training ', N, ' New Items ...')
new_data_file= "new_data.csv"
old_data_file= "ServerModel/uci-news-aggregator.csv"
new_weights_file= "new_weights.csv"
old_weights_file= "weights.csv"
def _normalize_text(s):
#lower case everything
s = s.lower()
#remove punctuation, exclude word related ones
s = re.sub('\s\W',' ',s)
s = re.sub('\W\s',' ',s)
#remove double spaces
s = re.sub('\s+',' ',s)
return s
#Optain new Training data
data = pd.read_csv(new_data_file)
data['TEXT'] = [_normalize_text(s) for s in data['TITLE']]
news= pd.read_csv(old_data_file)
news['TEXT'] = [_normalize_text(s) for s in news['TITLE']]
full_data=pd.concat([data, news])
tfidf1 = TfidfVectorizer()
# recaclulate tfidf
x = tfidf1.fit_transform(full_data['TEXT'])
tfidf = TfidfVectorizer(vocabulary=tfidf1.vocabulary_)
#fit the new data
x= tfidf.fit_transform(data['TEXT'])
encoder = LabelEncoder()
y = encoder.fit_transform(full_data['CATEGORY'])
#update weights
weights = np.genfromtxt( old_data_file ,delimiter=",",dtype=float)
Added_y= x.shape[1]-weights.shape[1]
added_weights= np.zeros((4,Added_y+1))
new_weights = np.append(weights,added_weights,1)
np.savetxt(new_weights_file, new_weights, delimiter=",")
log_reg= MultiClassLogisticRegression(y, weights_file= new_weights_file)
log_reg1= MultiClassLogisticRegression(y, weights_file= old_weights_file)
y_new = encoder.transform(data['CATEGORY'])
#Train 10 Patches of size N/10
l1=0
inc = int (N/10)
l2 = inc
for i in range(10):
print(l1,":",l2)
ss=x[l1:l2]
ss=ss.toarray()
log_reg.fit(ss,y[l1:l2],lr=0.003, verbose=True,weights_file=new_weights_file)
l1+=inc
l2+=inc
reset=np.array([["TITLE","CATEGORY"]])
np.savetxt(new_data_file,rest, delimiter=",") #remove trained data
print ("Training Completed!!")
self.test(tfidf,encoder,log_reg,log_reg1)
return