西瓜书-模型评估与选择Friedman检验python实现

Myth丶恋晨 2023-07-17 15:41 156阅读 0赞

冗长的定义。。  
计算Tf的函数在最底下，目前只实现了计算Tf，后续的有时间补上  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 1]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 2]  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 3]

import numpy as np
    data = np.array([[1,2,3],[1,2.5,2.5],[1,2,3],[1,2,3]])#书上的数据
    def Friedman(n,k,data_matrix):
        ''' Friedman检验 参数：数据集个数n, 算法种数k, 排序矩阵data_matrix 返回值是Tf '''
        
        #计算每个算法的平均序值，即求每一列的排序均值
        hang,lie = data_matrix.shape#查看数据形状
        XuZhi_mean = list()
        for i in range(lie):#计算平均序值
            XuZhi_mean.append(data_matrix[:,i].mean())
        
        sum_mean = np.array(XuZhi_mean)#转换数据结构方便下面运算
        ## 计算总的排序和即西伽马ri^2
        sum_ri2_mean = (sum_mean ** 2).sum()
        #计算Tf
        result_Tx2 = (12 * n) * (sum_ri2_mean - ((k * (k + 1) ** 2) / 4)) / (k * (k + 1))
        result_Tf = (n - 1) * result_Tx2 / (n * (k - 1) - result_Tx2)
        
        return result_Tf
    
    Tf = Friedman(4,3,data)
    print(Tf)

输出结果：  
![在这里插入图片描述][20200322220445683.png]  
![在这里插入图片描述][20200322220606163.png]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70]: /images/20230528/954e932d37b9471c9b01048ae9ce62cf.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 1]: /images/20230528/294629fb6d354466820ad6da296f558e.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 2]: /images/20230528/307c82b263584d09bf8d896a1fa72299.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2h1YW5namlhYWFhYQ_size_16_color_FFFFFF_t_70 3]: /images/20230528/f6b4b58fb52c4a81968544573ba6b3ea.png
[20200322220445683.png]: /images/20230528/359643c6ff94447a80412019bc4e4ca6.png
[20200322220606163.png]: /images/20230528/a5bf92a8e7a24888bbc7a093800a7d84.png