import pandas as pd
import numpy as np

df = pd.read_csv('rfm.csv')
df.head()
'''
    CustomerID  Recency Frequency   Monetary
0   12347.0     2        7          2866.77
1   12348.0     248      1          17.00
2   12349.0     18       1          1155.75
3   12350.0     310      1          274.00
4   12352.0     36       7          1147.44
'''


from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_score
import matplotlib.pyplot as plt

# 确定数据源
X = df.iloc[:,1:]

X.head()
'''
    Recency Frequency   Monetary
0   2        7          2866.77
1   248      1          17.00
2   18       1          1155.75
3   310      1          274.00
4   36       7          1147.44
'''


# 简单查看下如何创建模型
# 模型 = 算法 + 预处理后的数据
kmeans = KMeans(n_clusters=2,random_state=666,n_jobs=-1)
y_predict = kmeans.fit_predict(X)
silhouette_score(X,y_predict)
#结果：0.9727905413052251  此数接近1，说明构建的模型还是可行的


#查看模型的属性：labels_
kmeans.labels_
'''
array([[9.27459745e+01, 3.87262677e+00, 8.82704240e+02],
       [2.50000000e+00, 1.03000000e+02, 4.30544575e+04]])
'''

#重新训练新的数据:X
kmeans.predict(X)
'''
array([0, 0, 0, ..., 0, 0, 0], dtype=int32)
'''


# 创建学习曲线，查看查看轮廓系数,选择最合适的类别
# 用一个列表记录轮廓系数得分
s_list = []

'''
不确定分成多少类，可先试下分成 2-20个类别，查看一下轮廓系数
n_clusters
random_state
n_jobs
'''

for i in range(2,21):
    k = KMeans(n_clusters=i,random_state=666,n_jobs=-1)
    y_predict = k.fit_predict(X)
    
    score = silhouette_score(X,y_predict)
    s_list.append(score)
    
'''
绘制学习曲线
设置X轴间距
from matplotlib.pyplot import MultipleLocator
ax.xaxis.set_major_locator(MultipleLocator(1))
'''
from matplotlib.pyplot import MultipleLocator

figure,ax = plt.subplots()

ax.plot(range(2,21),s_list)
ax.xaxis.set_major_locator(MultipleLocator(1))

学习曲线1.png

# 继续寻找，在2-5之间，寻找最优得分
s_list = []

for i in range(2,6):
    k = KMeans(n_clusters=i,random_state=666,n_jobs=-1)
    y_predict = k.fit_predict(X)
    
    score = silhouette_score(X,y_predict)
    s_list.append(score)
    
''' 
从图示上看，最优的簇 是 2
不过一般而言，簇都会大于2，所以这里最优簇选择3
'''
from matplotlib.pyplot import MultipleLocator

figure,ax = plt.subplots()

ax.plot(range(2,6),s_list)
ax.xaxis.set_major_locator(MultipleLocator(1))

学习曲线2.png

# 用户簇类标签 
# labels_
k = KMeans(n_clusters=3,random_state=666,n_jobs=-1)
k.fit(X)
'''
KMeans(algorithm='auto', copy_x=True, init='k-means++', max_iter=300,
       n_clusters=3, n_init=10, n_jobs=-1, precompute_distances='auto',
       random_state=666, tol=0.0001, verbose=0)
'''

k.labels_
#结果：array([2, 0, 0, ..., 0, 0, 0], dtype=int32)

df['cluster'] = k.labels_
df.head()
'''

CustomerID  Recency Frequency   Monetary    cluster
0   12347.0 2        7          2866.77      2
1   12348.0 248      1          17.00        0
2   12349.0 18       1          1155.75      0
3   12350.0 310      1          274.00       0
4   12352.0 36       7          1147.44      0
'''