用于评估聚类效果的DBI实现

最近进行了一些聚类的学习，需要对聚类效果进行评估，用python进行了简单的DBI实现，给自己做一个小的记录。
DBI的理论知识见：http://blog.sina.com.cn/s/blog_65c8baf901016flh.html

```python
import math
import pandas as pd
import os
import numpy as np


cluster_path = os.getcwd() + '/1.csv'
data_path = os.getcwd() + '/数据.csv'
# read all data from csv
f = open(cluster_path)
cluster_data = pd.read_csv(f)
cluster_data_array = np.array(cluster_data)

f = open(data_path)
data_data = pd.read_csv(f)
data_data_array = np.array(data_data)


# print(cluster_data)
# print("---------------------------")
# print(cluster_data_array)


# 求两个点的欧式距离
def VD(v1, v2):
    s = 0
    for i in range(1, 9):
        s += (v1[i] - v2[i]) ** 2
    return s ** 0.5


# 求两个簇质心的距离
def Compute_decent(center1, center2):
    return VD(center1, center2)


# 求簇i内任两点距离的和
def Compute_Ci(x, clusterIndex):
    Ci = 0
    for i in range(x.shape[0] - 1):
        if x[i][0] == clusterIndex:
            for j in range(i + 1, x.shape[0]):
                if x[j][0] == clusterIndex:
                    Ci += VD(x[i], x[j])
    return Ci


# 取Ci的均值
def Ci_avg(x, ClusterIndex):
    clusterNum = 0
    for i in range(x.shape[0]):
        if x[i][0] == ClusterIndex:
            clusterNum += 1
    n = clusterNum * (clusterNum-1) / 2
    return Compute_Ci(x, ClusterIndex) / n


# 计算max后面那个式子
def Compute_Mij(avgCi, avgCj, denominator):
    numerator = avgCi + avgCj
    if denominator == 0:
        print('error！ denominator is 0')
    return numerator / denominator

# nc就是聚类数目，由于测试的数据量相对较大，用array把值存储下来，避免重复计算
def Compute_DBI(x, y, nc):
    # s--final result;
    # avgCi--1*nc list stores avgCi
    # decentList--nc*nc list stores decent
    s = 0
    avgCi = np.zeros(nc)
    decentList = np.zeros((nc, nc))

    for i in range(nc):
        for j in range(nc):
            decentList[i][j] = Compute_decent(y[i], y[j])
    print('decentList store over')

    for i in range(nc):
        avgCi[i] = Ci_avg(x, i + 1)
        print(avgCi[i])
        print('avgCi', i + 1, 'store over')

    for i in range(nc):
        print('epoc is:', i + 1)
        m = 0
        for j in range(i + 1, nc):
            temp = Compute_Mij(avgCi[i], avgCi[j], decentList[i][j])
            if m < temp:
                m = temp
        print('epoc', i + 1, 'over')
        s += m
    return s / nc


print(Compute_DBI(data_data_array, cluster_data_array, 12))