kaggle文本分类

import pandas as pd
from keras.layers import Dense,LSTM,Bidirectional,Embedding
from keras.models import Sequential
import keras.preprocessing as preprocessing
from nltk.corpus import stopwords
from keras.utils.np_utils import to_categorical
import nltk
import numpy as np
from keras.preprocessing.text import Tokenizer
stoplist = stopwords.words('english')
data = pd.read_csv(r'C:\Users\yuanchaoyi\Desktop\ka\train.tsv',sep='\t')
data_test = pd.read_csv(r'C:\Users\yuanchaoyi\Desktop\ka\test.tsv',sep='\t')
list = []
list_test = []
data_X = data['Phrase']
data_Y  = data['Sentiment']
data_test_X = data_test['Phrase']
[list.append(i) for i in data_X]
stop_data_X = []
for i in list:
    if str(i).lower().split(' ') not in stoplist:
        stop_data_X.append(i)
tokenizer = Tokenizer(num_words=10000)
tokenizer.fit_on_texts(stop_data_X)
X_train = tokenizer.texts_to_sequences(stop_data_X) #传入列表
X_train_data = preprocessing.sequence.pad_sequences(X_train,maxlen=300)
tokenizer = Tokenizer(num_words=10000)
tokenizer.fit_on_texts(data_test_X)
X_test = tokenizer.texts_to_sequences(data_test_X) #传入列表
X_test_data = preprocessing.sequence.pad_sequences(X_test,maxlen=300)
list_Y = []
[list_Y.append(i) for i in data_Y]
list_Y = np.array(list_Y)
Y_train = to_categorical(list_Y) #向量化标签
model = Sequential()
model.add(Embedding(10000,32,input_length=300))
model.add(LSTM(32))
model.add(Dense(5,activation='softmax'))
model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['acc'])
model.fit(X_train_data,Y_train,batch_size=32,epochs=5,validation_split=0.2)
a = model.predict(X_test_data)
print(a)