文本分类实战
本次采用的数据集分别是亚马逊商品评论数据（amazon_cells_labelled.txt）、IMDB电影评论数据(imdb_labelled.txt)、Yelp网站点评数据(yelp_labelled.txt)。数据下载地址为：
0. 读取数据
import pandas as pd file_dict = { 'amazon': 'amazon_cells_labelled.txt', 'imdb': 'imdb_labelled.txt', 'yelp':'yelp_labelled.txt' } total_df = pd.DataFrame() for k, v in file_dict.items(): single_df = pd.read_csv(v, sep = '\t', names = ['sent', 'label']) single_df['source'] = k total_df = total_df.append(single_df) total_df = total_df.dropna() #取出所有只要有NA的行 1. baseline模型
每次都先以Yelp数据集为例。
1.1 词袋模型 + 逻辑回归
from sklearn.feature_extraction.text import CountVectorizer from sklearn.model_selection import train_test_split from sklearn.linear_model import LogisticRegression yelp = total_df[total_df.source == 'yelp'] x = yelp['sent'] y = yelp['label'] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer.fit(x_train) x_train = vectorizer.transform(x_train) x_test = vectorizer.transform(x_test) logisitic_regression = LogisticRegression() logisitic_regression.fit(x_train, y_train) logisitic_regression.score(x_test, y_test) # 0.796 def get_df_predict_score(df): df = df[df['label'].notnull()] x = df['sent'] y = df['label'] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer = CountVectorizer() vectorizer.fit(x_train) x_train = vectorizer.transform(x_train).toarray() x_test = vectorizer.transform(x_test).toarray() logisitic_regression = LogisticRegression() logisitic_regression.fit(x_train, y_train) score = logisitic_regression.score(x_test, y_test) return score for i in file_dict.keys(): score = get_df_predict_score(total_df[total_df['source'] == i]) print('{} data score is {}'.format(i, score)) 1.2 TF-IDF模型 + 逻辑回归
from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.model_selection import train_test_split from sklearn.linear_model import LogisticRegression yelp = total_df[total_df.source == 'yelp'] x = yelp['sent'] y = yelp['label'] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer.fit(x_train) x_train = vectorizer.transform(x_train) x_test = vectorizer.transform(x_test) logisitic_regression = LogisticRegression() logisitic_regression.fit(x_train, y_train) logisitic_regression.score(x_test, y_test) # 0.796 def get_df_predict_score(df): df = df[df['label'].notnull()] x = df['sent'] y = df['label'] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer = TfidfVectorizer() vectorizer.fit(x_train) x_train = vectorizer.transform(x_train).toarray() x_test = vectorizer.transform(x_test).toarray() logisitic_regression = LogisticRegression() logisitic_regression.fit(x_train, y_train) score = logisitic_regression.score(x_test, y_test) return score for i in file_dict.keys(): score = get_df_predict_score(total_df[total_df['source'] == i]) print('{} data score is {}'.format(i, score)) 2. 深度学习模型
在深度学习中，通过观察训练过程中loss和metrics（在分类问题中可能为准确率）的变化，才能对模型的质量有个大体的评估。在Keras中，绘制loss和准确率曲线的代码如下所示：
其中plt.legend()是显示图例，如下图所示：
import matplotlib.pyplot as plt def plot_history(history, name): train_acc = history.history['acc'] train_loss = history.history['loss'] val_acc = history.history['val_acc'] val_loss = history.history['val_loss'] x = range(1, len(train_acc) + 1) plt.figure(figsize=(12, 5)) plt.subplot(1, 2, 1) plt.plot(x, train_acc, 'b', label='Training acc') plt.plot(x, val_acc, 'r', label='Validation acc') plt.title('Training and Validation accuracy') plt.legend() plt.subplot(1, 2, 2) plt.plot(x, train_loss, 'b', label='Training loss') plt.plot(x, val_loss, 'r', label='Validation loss') plt.title('Training and Validation loss') plt.legend() plt.show() plt.savefig(str(name) + '.png') 2.1 词袋模型 + MLP
from sklearn.feature_extraction.text import CountVectorizer import keras from keras.models import Sequential from keras.layers import Dense, Activation from sklearn.model_selection import train_test_split x = yelp['sent'] y = yelp['label'] #划分训练集和测试集 x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer = CountVectorizer() vectorizer.fit(x_train) x_train = vectorizer.transform(x_train).toarray() x_test = vectorizer.transform(x_test).toarray() input_dim = x_train.shape[1] model = Sequential() model.add(Dense(units = 10, input_dim = input_dim)) model.add(Dense(units = 1, activation = 'sigmoid')) model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy']) history = model.fit(x_train, y_train, batch_size = 32, epochs = 20) loss, accuracy = model.evaluate(x_test, y_test) print('test accuracy is', accuracy) import keras from keras.layers import Dense from keras.models import Sequential from sklearn.feature_extraction.text import CountVectorizer from sklearn.model_selection import train_test_split def get_df_predict_score(df): df = df[df['label'].notnull()] x = df['sent'] y = df['label'] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.25, random_state = 1000) vectorizer = CountVectorizer() vectorizer.fit(x_train) x_train = vectorizer.transform(x_train).toarray() x_test = vectorizer.transform(x_test).toarray() input_dim = x_train.shape[1] model = Sequential() model.add(Dense(input_dim = input_dim, units = 10, activation="relu")) model.add(Dense(units = 1, activation = 'sigmoid')) model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy']) history = model.fit(x_train, y_train, epochs = 20, batch_size = 32, validation_split=0.2, verbose = False) plot_history(history, 'result') loss, accuracy = model.evaluate(x_test, y_test) return accuracy for i in file_dict.keys(): score = get_df_predict_score(total_df[total_df['source'] == i]) print('{} data score is {}'.format(i, score)) 
