-
Notifications
You must be signed in to change notification settings - Fork 0
/
preprocessing_lstm.py
137 lines (123 loc) · 5.62 KB
/
preprocessing_lstm.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
import os
import re
import csv
import codecs
# data
import numpy as np
import pandas as pd
# text preprocessing
from string import punctuation
from collections import defaultdict
from nltk.corpus import stopwords
from nltk.stem import SnowballStemmer
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
from sklearn.preprocessing import StandardScaler
MAX_NB_WORDS = 200000
MAX_SEQUENCE_LENGTH = 40
VALIDATION_SPLIT = 0.1
GLOVE_DIR = './'
EMBEDDING_DIM = 300
def tokenization(texts_1, texts_2, test_texts_1, test_texts_2):
# tokenize
tokenizer = Tokenizer(nb_words=MAX_NB_WORDS)
tokenizer.fit_on_texts(texts_1 + texts_2 + test_texts_1 + test_texts_2)
sequences_1 = tokenizer.texts_to_sequences(texts_1)
sequences_2 = tokenizer.texts_to_sequences(texts_2)
test_sequences_1 = tokenizer.texts_to_sequences(test_texts_1)
test_sequences_2 = tokenizer.texts_to_sequences(test_texts_2)
word_index = tokenizer.word_index
return sequences_1, sequences_2, test_sequences_1, test_sequences_2, word_index
def preprocessing():
# preprocessing
print("-- Preprocessing Start")
df_train = pd.read_csv('train.csv', header=None)
df_train = df_train.fillna(' ')
df_train.columns = ['id', 'id1', 'id2', 'question1', 'question2', 'is_duplicate']
texts_1 = df_train["question1"].values.tolist()
texts_2 = df_train["question2"].values.tolist()
labels = df_train['is_duplicate'].apply(int)
labels = np.array(labels)
print("-- Preprocessing - train completed")
df_test = pd.read_csv('test.csv', header=None)
df_test = df_test.fillna(' ')
df_test.columns = ['id', 'id1', 'id2', 'question1', 'question2']
test_texts_1 = df_test["question1"].values.tolist()
test_texts_2 = df_test["question2"].values.tolist()
test_ids = df_test["id"]
print("-- Preprocessing - test completed")
print("-- Preprocessing - tokenization")
sequences_1, sequences_2, test_sequences_1, test_sequences_2, word_index = tokenization(texts_1, texts_2, test_texts_1, test_texts_2)
print("-- Preprocessing - Sentence Truncate")
data_1 = pad_sequences(sequences_1, maxlen=MAX_SEQUENCE_LENGTH) #padding : add values at the end to compare phrases from different lengths
data_2 = pad_sequences(sequences_2, maxlen=MAX_SEQUENCE_LENGTH)
test_data_1 = pad_sequences(test_sequences_1, maxlen=MAX_SEQUENCE_LENGTH)
test_data_2 = pad_sequences(test_sequences_2, maxlen=MAX_SEQUENCE_LENGTH)
test_ids = np.array(test_ids)
return data_1, data_2, test_data_1, test_data_2, test_ids, labels, word_index
def load_leaky():
# new features
print("-- Preprocessing - load leaky features")
trn = pd.read_csv('X_train.csv', index_col=0)
trn = trn.drop(["is_duplicate","question1", "question2"], axis=1) #,"question1_nouns", "question2_nouns"
tst = pd.read_csv('X_test.csv', index_col=0)
tst = tst.drop(["question1", "question2"], axis=1) #,"question1_nouns", "question2_nouns"
trn = trn.replace([np.inf, -np.inf], np.nan)
tst = tst.replace([np.inf, -np.inf], np.nan)
trn = trn.fillna(value=0)
tst = tst.fillna(value=0)
trn.shape, tst.shape
print(trn.isnull().values.any())
print(tst.isnull().values.any())
print("-- Preprocessing - rescale leaky features")
leaks = trn[trn.columns.values]
test_leaks = tst[tst.columns.values]
ss = StandardScaler()
ss.fit(np.vstack((leaks, test_leaks)))
leaks = ss.transform(leaks)
test_leaks = ss.transform(test_leaks)
return leaks, test_leaks
def input_nn_data(data_1, data_2, test_data_1, test_data_2, test_ids, labels, leaks, test_leaks):
np.random.seed(1234)
perm = np.random.permutation(len(data_1))
idx_train = perm[:int(len(data_1)*(1-VALIDATION_SPLIT))]
idx_val = perm[int(len(data_1)*(1-VALIDATION_SPLIT)):]
data = {}
data["data_1_train"] = np.vstack((data_1[idx_train], data_2[idx_train]))
data["data_2_train"] = np.vstack((data_2[idx_train], data_1[idx_train]))
data["leaks_train"] = np.vstack((leaks[idx_train], leaks[idx_train]))
data["labels_train"] = np.concatenate((labels[idx_train], labels[idx_train]))
data["data_1_val"] = np.vstack((data_1[idx_val], data_2[idx_val]))
data["data_2_val"] = np.vstack((data_2[idx_val], data_1[idx_val]))
data["leaks_val"] = np.vstack((leaks[idx_val], leaks[idx_val]))
data["labels_val"] = np.concatenate((labels[idx_val], labels[idx_val]))
data["weight_val"] = np.ones(len(data["labels_val"]))
test = pd.read_csv('test.csv')
test = test.fillna(' ')
test.columns = ['test_id', 'id1', 'id2', 'question1', 'question2']
data["test"] = test
return data
def Glove_Indexing():
# Indexing Glove
print('Indexing word vectors.')
embeddings_index = {}
with codecs.open(os.path.join(GLOVE_DIR, 'glove.840B.300d.txt'), encoding='utf-8') as f:
for line in f:
values = line.split(' ')
word = values[0]
coefs = np.asarray(values[1:], dtype='float32')
embeddings_index[word] = coefs
return embeddings_index
def Words_Embedding(word_index, embeddings_index):
# Embeddings
print('Preparing embedding matrix')
nb_words = min(MAX_NB_WORDS, len(word_index))+1
embedding_matrix = np.zeros((nb_words, EMBEDDING_DIM))
for word, i in word_index.items():
embedding_vector = embeddings_index.get(word)
if embedding_vector is not None:
embedding_matrix[i] = embedding_vector
else:
embedding_matrix[i] = np.random.normal(scale=0.6, size=(EMBEDDING_DIM,))
print('Null word embeddings: %d' % np.sum(np.sum(embedding_matrix, axis=1) == 0))
return embedding_matrix