机器学习算法之聚类算法拓展：K-Means和Mini Batch K-Means算法效果评估

聚类算法的衡量指标

混淆矩阵
均一性
完整性
V-measure
调整兰德系数(ARI)
调整互信息(AMI)
轮廓系数(Silhouette)

import time
import numpy as np  
import matplotlib.pyplot as plt  
import matplotlib as mpl
from sklearn.cluster import MiniBatchKMeans, KMeans 
from sklearn import metrics
from sklearn.metrics.pairwise import pairwise_distances_argmin  
from sklearn.datasets.samples_generator import make_blobs 

## 设置属性防止中文乱码
mpl.rcParams['font.sans-serif'] = [u'SimHei']
mpl.rcParams['axes.unicode_minus'] = False

centers = [[1, 1], [-1, -1], [1, -1]] 
clusters = len(centers)       

X, Y = make_blobs(n_samples=3000, centers=centers, cluster_std=0.7, random_state=28) 
Y # 在实际工作中是人工给定的，专门用于判断聚类的效果的一个值

array([2, 0, 0, …, 2, 2, 1])

k_means = KMeans(init='k-means++', n_clusters=clusters, random_state=28)
t0 = time.time() 
k_means.fit(X)  
km_batch = time.time() - t0  
print ("K-Means算法模型训练消耗时间:%.4fs" % km_batch)

K-Means算法模型训练消耗时间:0.1211s

batch_size = 100
mbk = MiniBatchKMeans(init='k-means++', n_clusters=clusters, batch_size=batch_size, random_state=28)  
t0 = time.time()  
mbk.fit(X)  
mbk_batch = time.time() - t0  
print ("Mini Batch K-Means算法模型训练消耗时间:%.4fs" % mbk_batch)

Mini Batch K-Means算法模型训练消耗时间:0.0991s

km_y_hat = k_means.labels_
mbkm_y_hat = mbk.labels_
print(km_y_hat) # 样本所属的类别

[0 2 2 … 1 1 0]

k_means_cluster_centers = k_means.cluster_centers_
mbk_means_cluster_centers = mbk.cluster_centers_
print ("K-Means算法聚类中心点:\ncenter=", k_means_cluster_centers)
print ("Mini Batch K-Means算法聚类中心点:\ncenter=", mbk_means_cluster_centers)
order = pairwise_distances_argmin(k_means_cluster_centers,  
                                  mbk_means_cluster_centers) 
order

K-Means算法聚类中心点:
center= [[-1.0600799 -1.05662982]
[ 1.02975208 -1.07435837]
[ 1.01491055 1.02216649]]
Mini Batch K-Means算法聚类中心点:
center= [[ 0.99602094 1.10688195]
[-1.00828286 -1.05983915]
[ 1.07892315 -0.94286826]]
array([1, 2, 0], dtype=int64)

### 效果评估
score_funcs = [
    metrics.adjusted_rand_score,#ARI
    metrics.v_measure_score,#均一性和完整性的加权平均
    metrics.adjusted_mutual_info_score,#AMI
    metrics.mutual_info_score,#互信息
]

## 2. 迭代对每个评估函数进行评估操作
for score_func in score_funcs:
    t0 = time.time()
    km_scores = score_func(Y,km_y_hat)
    print("K-Means算法:%s评估函数计算结果值:%.5f；计算消耗时间:%0.3fs" % (score_func.__name__,km_scores, time.time() - t0))
    
    t0 = time.time()
    mbkm_scores = score_func(Y,mbkm_y_hat)
    print("Mini Batch K-Means算法:%s评估函数计算结果值:%.5f；计算消耗时间:%0.3fs\n" % (score_func.__name__,mbkm_scores, time.time() - t0))

K-Means算法:adjusted_rand_score评估函数计算结果值:0.72526；计算消耗时间:0.223s
Mini Batch K-Means算法:adjusted_rand_score评估函数计算结果值:0.72421；计算消耗时间:0.002s

K-Means算法:v_measure_score评估函数计算结果值:0.65754；计算消耗时间:0.013s
Mini Batch K-Means算法:v_measure_score评估函数计算结果值:0.65780；计算消耗时间:0.002s

K-Means算法:adjusted_mutual_info_score评估函数计算结果值:0.65726；计算消耗时间:0.031s
Mini Batch K-Means算法:adjusted_mutual_info_score评估函数计算结果值:0.65757；计算消耗时间:0.004s

K-Means算法:mutual_info_score评估函数计算结果值:0.72231；计算消耗时间:0.002s
Mini Batch K-Means算法:mutual_info_score评估函数计算结果值:0.72264；计算消耗时间:0.002s