基于收支数据的聚类分析研究

聚类方法仍需要对分群结果进行解读&＃xff0c;通过业务合理性来选择分群的数量

import pandas as pd
import matplotlib.pyplot as plt


dataset&＃61;pd.read_csv("customers.csv")


dataset.head()CustomerID Genre Age Annual Income (k$) Spending Score (1-100)
0 1 Male 19 15 39
1 2 Male 21 15 81
2 3 Female 20 16 6
3 4 Female 23 16 77
4 5 Female 31 17 40


dataset.describe()CustomerID Age Annual Income (k$) Spending Score (1-100)
count 200.000000 200.000000 200.000000 200.000000
mean 100.500000 38.850000 60.560000 50.200000
std 57.879185 13.969007 26.264721 25.823522
min 1.000000 18.000000 15.000000 1.000000
25% 50.750000 28.750000 41.500000 34.750000
50% 100.500000 36.000000 61.500000 50.000000
75% 150.250000 49.000000 78.000000 73.000000
max 200.000000 70.000000 137.000000 99.000000


X &＃61; dataset.iloc[:, [3, 4]].values
#全部行&＃xff0c;第四第五列 Annual Income (k$) 和 Spending Score (1-100)


from sklearn.cluster import KMeans
kmeans &＃61; KMeans(n_clusters &＃61; 5, init &＃61; &＃39;k-means&＃43;&＃43;&＃39;, random_state &＃61; 42)#k&＃61;5
# random_state &＃61; 42设置完之后&＃xff0c;建模生成的随机数都是一样的
# n_clusters分成多少组
# Kmeans随机选几个点&＃xff0c;然后开始算距离&＃xff0c;距离离得近的属于一类&＃xff0c;给每个类别打上标签&＃xff0c;算出聚类中心。离的很近或者小于阈值&＃xff0c;聚类结束。
# Kmean&＃43;&＃43; 第一个随机选择。n&＃43;1是选择离第一个远的距离。
y_kmeans &＃61; kmeans.fit_predict(X) #每个记录x都给预测了一个聚类的值


plt.scatter(X[y_kmeans &＃61;&＃61; 0, 0], X[y_kmeans &＃61;&＃61; 0, 1], s &＃61; 100, c &＃61; &＃39;red&＃39;, label &＃61; &＃39;Standard&＃39;)
plt.scatter(X[y_kmeans &＃61;&＃61; 1, 0], X[y_kmeans &＃61;&＃61; 1, 1], s &＃61; 100, c &＃61; &＃39;blue&＃39;, label &＃61; &＃39;Traditional&＃39;)
plt.scatter(X[y_kmeans &＃61;&＃61; 2, 0], X[y_kmeans &＃61;&＃61; 2, 1], s &＃61; 100, c &＃61; &＃39;green&＃39;, label &＃61; &＃39;Normal&＃39;)
plt.scatter(X[y_kmeans &＃61;&＃61; 3, 0], X[y_kmeans &＃61;&＃61; 3, 1], s &＃61; 100, c &＃61; &＃39;cyan&＃39;, label &＃61; &＃39;Youth&＃39;)
plt.scatter(X[y_kmeans &＃61;&＃61; 4, 0], X[y_kmeans &＃61;&＃61; 4, 1], s &＃61; 100, c &＃61; &＃39;magenta&＃39;, label &＃61; &＃39;TA&＃39;)
#分成五类&＃xff0c;分别把值选出来。0,1,2,3,4,5代表5类&＃xff1b;0,1代表数字x的值&＃xff1b;s点的大小。c是颜色&＃xff0c;label是标签
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s &＃61; 300, c &＃61; &＃39;black&＃39;, label &＃61; &＃39;Centroids&＃39;)
#cluster_centers聚类中心
plt.title(&＃39;Clusters of customers&＃39;)
plt.xlabel(&＃39;Annual Income (k$)&＃39;)
plt.ylabel(&＃39;Spending Score (1-100)&＃39;)
plt.legend()
plt.show()


import matplotlib.pyplot as pltwcss &＃61; []for i in range(1, 11): #循环使用不同k测试结果kmeans &＃61; KMeans(n_clusters &＃61; i, init &＃61; &＃39;k-means&＃43;&＃43;&＃39;, random_state &＃61; 42)kmeans.fit(X)wcss.append(kmeans.inertia_) #inertia簇内误差平方和plt.plot(range(1, 11), wcss)plt.title(&＃39;The Elbow Method&＃39;)plt.xlabel(&＃39;Number of clusters&＃39;)plt.ylabel(&＃39;WCSS&＃39;)plt.show()
#内核的数量应该怎么选比较合适&＃xff0c;在课堂中是尝试 10 个内核&＃xff0c;取值是1到11。反正就是多测试几次&＃xff0c;出现最佳正确答案就可以了。取得这个值没有定数。


结论: 5个分群比较好