pyLDAvis 模块代码及应用

背景

pyLDAvis模块是python中的一个对LDA主题模型算法的可视化模块。本文的代码是根据github上的某个项目代码修改而得，很感谢github及创造原始代码的大牛朋友们！

import pandas as pd

df = pd.read_csv("C:\\Users\\Desktop\\neg.csv",errors='ignore')

print(df.head())

print(df.shape)

import jieba

jieba.load_userdict("C:\\Users\\Desktop\\中文分词词库整理\\中文分词词库整理\\百度分词词库.txt") #自定义分词词典

def chinese_word_cut(mytext):

    return " ".join(jieba.cut(mytext))#分词

df["content_cutted"] = df.content.apply(chinese_word_cut)

print(df.content_cutted.head())

from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer

n_features = 1000

tf_vectorizer = CountVectorizer(strip_accents = 'unicode',

                                max_features=n_features,

                                stop_words='english',

                                max_df = 1.0,

                                min_df = 0.1)#训练词矩阵

tf = tf_vectorizer.fit_transform(df.content_cutted)

from sklearn.decomposition import LatentDirichletAllocation

n_topics = 3

lda = LatentDirichletAllocation(n_topics=n_topics, max_iter=50,

                                learning_method='online',

                                learning_offset=50.,

                                random_state=0)#LDA模型训练

lda.fit(tf)

def print_top_words(model, feature_names, n_top_words):#主题相关的top词计算

    for topic_idx, topic in enumerate(model.components_):

        print("Topic #%d:" % topic_idx)

        print(" ".join([feature_names[i]

            for i in topic.argsort()[:-n_top_words - 1:-1]]))

    print()

n_top_words = 25

tf_feature_names = tf_vectorizer.get_feature_names()

print_top_words(lda, tf_feature_names, n_top_words)

import pyLDAvis#所需可视化模块

import pyLDAvis.sklearn

data = pyLDAvis.sklearn.prepare(lda, tf, tf_vectorizer)

pyLDAvis.show(data)#可视化主题模型