Question

我建立了一个LDA模型和术语doc-matrix。现在，我想在'df ['topic-description']'中添加主题。我已经为此编写了一个函数，但是代码似乎有些错误。我查看了idxmax对对象类型不起作用的stackoverflow，因此将dtype更改为数字。我无法在这里找出问题。感谢您的帮助

向df添加主题的功能

def add_topics_to_df(ldamodel, doc_term_matrix, df, new_col='topics_description', num_topics = 3):
    # Convert into Per-document topic probability matrix:
    docTopicProbMat = ldamodel[doc_term_matrix]
    docTopicProbDf = pd.DataFrame(index=df.index, columns=range(0, num_topics))
    for i, doc in enumerate(docTopicProbMat):
        for topic in doc:
            docTopicProbDf.iloc[i, topic[0]] = topic[1]
    docTopicProbDf[new_col] = pd.to_numeric(docTopicProbDf[new_col])
    docTopicProbDf[new_col] = docTopicProbDf.idxmax(axis=1)
    df_topics = docTopicProbDf[new_col]
    # Merge with df
    df_new = pd.concat([df, df_topics], axis=1)
    return df_new

函数调用和进一步处理


df = add_topics_to_df(ldamodel_description, doc_term_matrix_description, df, new_col='topics_description', num_topics=3)

# Rename based on understanding of topics
df['topics_description'].replace({0:'Location', 1:'Luxury', 2:'Budget'},inplace=True)
df = pd.get_dummies(df, columns=['topics_description'], drop_first=False)

print("Dataset has {} rows, {} columns.".format(*df.shape))

将主题添加为df中的功能

0 个答案: