python - matplotlib 中的并排词云

Question

我正在使用包 WordCloud显示由 scikit LDA 生成的单词(潜在狄利克雷分配)。对于 LDA 生成的每个主题，我都会有一个图表。我希望能够在网格中绘制所有图表，以便并排可视化。本质上，我有一个将 LDA 模型作为输入的函数，以及我想要可视化的 LDA 主题，然后绘制一个词云:

from wordcloud import WordCloud
import matplotlib.pyplot as plt
SEED=0

def topicWordCloud(model, topicNumber, WCmaxWords,WCwidth, WCheight):
    topic = model.components_[topicNumber]
    tupleList = [(tf_feature_names[i],int(topic[i]/topic.sum()*10000)) for i in range(len(topic))]
    wordcloud = WordCloud(width=WCwidth, height=WCheight, max_words=WCmaxWords, random_state=42).generate_from_frequencies(tupleList)
    plt.figure( figsize=(20,10) )
    plt.imshow(wordcloud)
    plt.axis("off")

topicWordCloud(model=lda, topicNumber=2, WCmaxWords=100,WCwidth=800, WCheight=600)

如何遍历所有主题 (n_topics) 以可视化网格中的所有图表？我在想一些事情:

fig = plt.figure()
for i in range(n_topics):
    plt.subplot(2,1,i+1) 
    #something here

Answer 1

从您的函数返回 wordcloud，然后从您的 for 循环中调用 topicWordCloud。然后，在使用 fig.add_subplot 创建的 Axes 上使用 imshow。例如，像这样:

def topicWordCloud(model, topicNumber, WCmaxWords,WCwidth, WCheight):
    topic = model.components_[topicNumber]
    tupleList = [(tf_feature_names[i],int(topic[i]/topic.sum()*10000)) for i in range(len(topic))]
    wordcloud = WordCloud(width=WCwidth, height=WCheight, max_words=WCmaxWords, random_state=42).generate_from_frequencies(tupleList)
    return wordcloud

fig = plt.figure()
for i in range(n_topics):
    ax = fig.add_subplot(2,1,i+1)
    wordcloud = topicWordCloud(...)

    ax.imshow(wordcloud)
    ax.axis('off')