【无标题】

greeneea

于 2024-09-24 00:42:30 发布

阅读量216

点赞数 1

文章标签：机器学习人工智能

本文链接：https://blog.csdn.net/greeneea/article/details/142471991

版权

 # 计算轮廓系数，评估聚类结果 
 dbscan的方法，到计算轮廓系数的地方都报错，不知道为啥

raise ValueError(
ValueError: Number of labels is 1. Valid values are 2 to n_samples - 1 (inclusive)

# k-means分类
def kmeans_sample():
    import pandas as pd
    from sklearn.feature_extraction.text import TfidfVectorizer   #向量化
    from sklearn.cluster import KMeans  #kmeans聚类方法
    from sklearn.metrics import silhouette_score  #

    # 示例数据
    data = {'新闻发布时间':['2024-01-01','2024-01-02','2024-01-03'],
            '网站':['site1','site2','site3'],
            '主体':['主体A','主体B','主体C'],
            '内容摘要': ['摘要1', '摘要2', '摘要3'],
            '甲方名': ['甲方1', '甲方2', '甲方3'],
            '乙方名': ['乙方1', '乙方2', '乙方3'],
            '协议内容':['协议1','协议2','协议3']
            }
    df = pd.DataFrame(data)
    # 使用TF-IDF向量化内容摘要
    tfidf_vectorizer = TfidfVectorizer(stop_words='english')
    tfidf_matrix = tfidf_vectorizer.fit_transform(df['内容摘要'])

    # 选择聚类的数量
    n_clusters = 2 #可根据需要调整
    kmeans = KMeans(n_clusters= n_clusters,random_state=42)
    kmeans.fit(tfidf_matrix)

    # 将聚类结果添加到原始DataFrame中
    df['cluster'] = kmeans.labels_

    # 计算轮廓系数，评估聚类结果
    # score = silhouette_score(tfidf_matrix,kmeans.labels_)
    # print(f'轮廓系数:{score}')

    # 将聚类结果保存到新的excel文件中
    output_filename = 'clustered_data.xlsx'
    with pd.ExcelWriter(output_filename,engine='openpyxl')as writer:
        df.to_excel(writer,index=False)
    print(f'聚类结果已保存到 {output_filename}')


# dbscan分类
def dbscan_sample():
    import pandas as pd
    from sklearn.feature_extraction.text import TfidfVectorizer  # 向量化
    from sklearn.cluster import DBSCAN  # DBSCAN聚类方法
    from sklearn.metrics import silhouette_score  #

    # 示例数据
    data = {'新闻发布时间': ['2024-01-01', '2024-01-02', '2024-01-03'],
            '网站': ['site1', 'site2', 'site3'],
            '主体': ['主体A', '主体B', '主体C'],
            '内容摘要': ['摘要1', '摘要2', '摘要3'],
            '甲方名': ['甲方1', '甲方2', '甲方3'],
            '乙方名': ['乙方1', '乙方2', '乙方3'],
            '协议内容': ['协议1', '协议2', '协议3']
            }
    df = pd.DataFrame(data)
    # 使用TF-IDF向量化内容摘要
    tfidf_vectorizer = TfidfVectorizer(stop_words='english')
    tfidf_matrix = tfidf_vectorizer.fit_transform(df['内容摘要'])

    # 选择聚类的数量
    # n_clusters = 2
    dbscan = DBSCAN(eps= 0.5,min_samples=2)# 可根据数据调整
    clusters = dbscan.fit_predict(tfidf_matrix)

    # 将聚类结果添加到原始DataFrame中
    df['cluster'] = clusters

    # 计算轮廓系数，评估聚类结果
    # score = silhouette_score(tfidf_matrix, clusters)
    # print(f'轮廓系数:{score}')

    # 将聚类结果保存到新的excel文件中
    output_filename = 'clustered_data_dbscan.xlsx'
    with pd.ExcelWriter(output_filename, engine='openpyxl')as writer:
        df.to_excel(writer, index=False)

    print(f'聚类结果已保存到 {output_filename}')