-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathNLPPipeline.py
executable file
·133 lines (117 loc) · 5.65 KB
/
NLPPipeline.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
import numpy as np
import pandas as pd
from sklearn.base import BaseEstimator
from sklearn.model_selection import train_test_split
from TextPreprocess import TextPreprocess
from DataSplit import DataSplit
from WordEmbedProcess import WordEmbedProcess
from DocClassifier import DocClassifier
'''
ycol = config['Ycol'],
xcol = config['Xcol'],
preprocs=config['analyticSettings']['preprocs'],
max_features=config['analyticSettings']['max_features'],
max_len=config['analyticSettings']['max_len'],
embed_size=config['analyticSettings']['max_len'],
train_batch_size=config['fitSettings']['train_batch_size'],
test_batch_size=config['fitSettings']['test_batch_size'],
epochs=config['fitSettings']['epochs'],
callbacks=config['fitSettings']['callbacks'],
best_model_path=config['fitSettings']['best_model_path']
'''
class NLPPipline(BaseEstimator):
def __init__(
self,
ycol,
xcol,
valid_size,
preWordEmbedPath,
max_features,
max_len,
embed_size,
train_batch_size,
test_batch_size,
epochs,
callbacks=None,
splittype=["train_test"],
preprocs=["raw", "fillna", "comment_to_lower", "text_to_seq"],
best_model_path="best_model.hdf5"
):
super(NLPPipline, self).__init__()
self.ycol = ycol
self.xcol = xcol
self.max_len = max_len
self.valid_size = valid_size
self.splittype = splittype
self.train_batch_size = train_batch_size
self.test_batch_size = test_batch_size
self.epochs = epochs
self.callbacks = callbacks
self.best_model_path = best_model_path
self.textPreprocess = TextPreprocess(preprocs=preprocs, max_features = max_features, max_len = max_len)
self.dataSplit = DataSplit(valid_size=valid_size, splittype=splittype)
self.preWordEmbedPath = preWordEmbedPath
self.wordEmbedProcess = WordEmbedProcess(max_features = max_features, embed_size = embed_size)
def get_pickable(self):
return {
'textPreprocess': self.textPreprocess.get_pickable(),
'dataSplit': self.dataSplit.get_pickable(),
'wordEmbedProcess': self.wordEmbedProcess.get_pickable(),
'docClassifier': self.docClassifier.get_pickable(),
'ycol': self.ycol,
'xcol': self.xcol
}
def load_pickable(self, pkl):
self.textPreprocess.load_pickable(pkl['textPreprocess'])
self.dataSplit.load_pickable(pkl['dataSplit'])
self.wordEmbedProcess.load_pickable(pkl['wordEmbedProcess'])
self.docClassifier.load_pickable(pkl['docClassifier'])
self.ycol = pkl['ycol']
self.xcol = pkl['xcol']
def fit(self, train_df):
"""
send original train data before splitting
"""
y_train = train_df[self.ycol]
self.textPreprocess.fit(train_df)
x_train_pre = self.textPreprocess.transform(train_df)
x_train, x_valid, y_train, y_valid = self.dataSplit.transform(x_train_pre, y_train)
glove_wordembed = WordEmbedProcess.load_pretrainedwordembed(self.preWordEmbedPath) # Note staticmethod is called with classname (ie before initializing class)
embedding_matrix = self.wordEmbedProcess.fit_transform(pretrainedwordembed=glove_wordembed,
word_index=self.textPreprocess.word_index)
self.docClassifier = DocClassifier(embedding_matrix=embedding_matrix,
max_len=self.max_len,
train_batch_size=self.train_batch_size,
test_batch_size=self.test_batch_size,
epochs=self.epochs,
callbacks=self.callbacks,
best_model_path=self.best_model_path)
self.docClassifier.fit(x_train, x_valid, y_train, y_valid)
def predict(self, test_df):
x_test_pre = self.textPreprocess.transform(test_df)
y_test_pred = self.docClassifier.predict(x_test_pre)
return y_test_pred
if __name__ == '__main__':
train_df = pd.read_csv('./input/train.csv')
test_df = pd.read_csv('./input/test.csv')
#x = train_df[["comment_text"]]
#y = train_df[["toxic", "severe_toxic", "obscene", "threat", "insult", "identity_hate"]]
nLPPipline = NLPPipline(
ycol=["toxic", "severe_toxic", "obscene", "threat", "insult", "identity_hate"],
xcol=["comment_text"],
valid_size=0.1,
preWordEmbedPath="./input/glove6b50d/glove.6B.50d.txt",
splittype=["train_test"],
preprocs=["raw", "fillna", "comment_to_lower", "text_to_seq"],
max_features=100000,
max_len=150,
embed_size=50,
train_batch_size=256,
test_batch_size=1024,
epochs=1,
callbacks=True,
best_model_path="best_model.hdf5"
)
nLPPipline.fit(train_df)
y_test_pred = nLPPipline.predict(test_df)
nLPPipline.get_pickable()