【综合评价方法 变异系数权重法】指标权重确定方法之变异系数权重法

# -*- encoding=utf-8 -*-

import pandas as pd
import numpy as np


# 自定义归一化函数

def autoNorm(data):
    """ :param data: 列表 :return: 归一化列表 """
    arr = np.asarray(data)
    norm_list=[]
    for x in arr:
        x = round(float(x - np.min(arr)) / ((np.max(arr) - np.min(arr))+0.001),4)
        norm_list.append(x)

    return norm_list



# 自定义热度值计算函数

def get_hot_value(context_train_data):

    """ :param context_train_data:数据框 :return: 热度值分数 """


    # 求相关列均值与标准差

    context_train_mean = context_train_data.mean(axis=0)

    context_train_std = context_train_data.std(ddof=0)

    # 求变异系数

    context_train_cof_var = context_train_std/context_train_mean

    # 对变异系数求和

    sum_context_train_cof_var = context_train_cof_var.sum()

    # 得出权重

    context_train_wi = context_train_cof_var/sum_context_train_cof_var

    # 将权重转换为矩阵

    cof_var = np.mat(context_train_wi)

    # 将数据框转换为矩阵
    context_train_data = np.mat(context_train_data)

    # 权重跟自变量相乘
    last_hot_matrix = context_train_data * cof_var.T
    last_hot_matrix = pd.DataFrame(last_hot_matrix.T)

    # 累加求和得到总分
    last_hot_score =list(last_hot_matrix.apply(sum))


    # max-min 归一化

    last_hot_score_autoNorm=autoNorm(last_hot_score)


    # 部落的热度值映射成分数（0-100分）

    last_hot_score_result=[i*100 for i in last_hot_score_autoNorm]



    return last_hot_score_result








if __name__ == '__main__':

    # 读取数据
    # context_train_data=pd.DataFrame(np.arange(24).reshape(4,6),columns=['x1','x2','x3','x4','x5','x6'])


    data=pd.read_excel('C:\\Users\\xiaohu\\Desktop\\文本挖掘\\部落的热度算法\\data.xlsx')

    # 拿出所有自变量数据(不包含部落id,部落名称)

    context_train_data=data.iloc[:,2:16]

    print(context_train_data)

    # 调用热度值计算函数
    last_hot_score_result =get_hot_value(context_train_data)

    # 增加一列部落名

    context_train_data['tribe_name']=data['tribe_name']


    # 增加一列热度值
    context_train_data['tribe_hot_value']=last_hot_score_result

    # 然后对数据框按热度值从大到小排序

    result=context_train_data.sort_values(by = 'tribe_hot_value',axis = 0,ascending = False)

    result['rank']=range(1,len(result)+1)

    print(result)


    # 输出excel

    # 写出csv数据
    result.to_csv('C:\\Users\\xiaohu\\Desktop\\文本挖掘\\部落的热度算法\\result.csv', index=False)