聚类是非监督学习的一种算法，我们使用k-means聚类算法，实现客户细分，以及营销战略如何在实际业务中应用。
1.导入数据
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.ticker as ticker
import seaborn as sns
from sklearn.cluster import KMeans
data = pd.read_csv('./Mall_Customers.csv')
2.数据探索
data.head()
| CustomerID | Gender | Age | Annual Income (k$) | Spending Score (1-100) | |
| 0 | 1 | Male | 19 | 15 | 39 |
| 1 | 2 | Male | 21 | 15 | 81 |
| 2 | 3 | Female | 20 | 16 | 6 |
| 3 | 4 | Female | 23 | 16 | 77 |
| 4 | 5 | Female | 31 | 17 | 40 |
data.info()
RangeIndex: 200 entries, 0 to 199
Data columns (total 5 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 CustomerID 200 non-null int64
1 Gender 200 non-null object
2 Age 200 non-null int64
3 Annual Income (k$) 200 non-null int64
4 Spending Score (1-100) 200 non-null int64
dtypes: int64(4), object(1)
memory usage: 7.9+ KB
data.isnull().any()
CustomerID False
Gender False
Age False
Annual Income (k$) False
Spending Score (1-100) False
dtype: bool
data.describe()
| CustomerID | Age | Annual Income (k$) | 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 |
data[['Gender','CustomerID']].groupby('Gender').count()
| CustomerID |
|Gender||
|Female|112|
|Male|88|
gender = data['Gender'].value_counts()
labels = ['Female', 'Male']
colors = ['c', 'coral']
explode = [0, 0.05]
plt.figure(figsize=(8,8))
plt.title('Total of customers by gender', fontsize = 16, fontweight='bold')
plt.pie(gender, colors = colors, autopct = '%1.0f%%', labels = labels, explode = explode, startangle=90, textprops={'fontsize': 16})
plt.savefig('Total of customers by gender.png', bbox_inches = 'tight')
plt.show()
output_11_0.png
plt.figure(figsize=(16,6))
plt.subplot(1,2,1)
sns.distplot(data['Spending Score (1-100)'], color = 'green')
plt.title('Distribution of Spending Score')
plt.subplot(1,2,2)
sns.distplot(data['Annual Income (k$)'], color = 'green')
plt.title('Distribution of Annual Income (k$)')
plt.show()
output_12_0.png
sns.pairplot(data=data[['Spending Score (1-100)','Annual Income (k$)','Age']], diag_kind="kde")
plt.savefig('Distribution.png', bbox_inches = 'tight')
output_13_0.png
plt.figure(figsize=(8,6))
plt.title('Annual Income vs Spending Score', fontsize = 16, fontweight='bold')
plt.scatter(data['Annual Income (k$)'], data['Spending Score (1-100)'], color = 'indianred', edgecolors = 'crimson')
plt.xlabel('Annual Income', fontsize = 14)
plt.ylabel('Spending Score', fontsize = 14)
plt.savefig('Annual Income vs Spending Score.png', bbox_inches = 'tight')
plt.show()
output_14_0.png
3.模型开发
X1_Matrix = data.iloc[:, [2,4]].values # Age & Spending Score
X2_Matrix = data.iloc[:, [3,4]].values # Annual Income & Spending Score
inertias_1 = []
for i in range(1,20):
kmeans = KMeans(n_clusters=i, init='k-means++', max_iter=300, n_init=10,random_state=0)
kmeans.fit(X1_Matrix)
inertia = kmeans.inertia_
inertias_1.append(inertia)
print('For n_cluster =', i, 'The inertia is:', inertia)
For n_cluster = 1 The inertia is: 171535.5
For n_cluster = 2 The inertia is: 75949.15601023017
For n_cluster = 3 The inertia is: 45840.67661610867
For n_cluster = 4 The inertia is: 28165.58356662934
For n_cluster = 5 The inertia is: 23830.24505228459
For n_cluster = 6 The inertia is: 19502.407839362204
For n_cluster = 7 The inertia is: 15523.684014328752
For n_cluster = 8 The inertia is: 13020.084512948222
For n_cluster = 9 The inertia is: 11517.231348351697
For n_cluster = 10 The inertia is: 10299.698359250398
For n_cluster = 11 The inertia is: 9404.802904325206
For n_cluster = 12 The inertia is: 8659.542579270144
For n_cluster = 13 The inertia is: 7896.277200074606
For n_cluster = 14 The inertia is: 7223.8088214073505
For n_cluster = 15 The inertia is: 6691.75644045497
For n_cluster = 16 The inertia is: 6160.592835350923
For n_cluster = 17 The inertia is: 5552.953625949214
For n_cluster = 18 The inertia is: 5356.265766259883
For n_cluster = 19 The inertia is: 4869.198509239299
# Creating the figure
figure = plt.figure(1, figsize=(15,6), dpi=300)
plt.plot(np.arange(1,20), inertias_1, alpha=0.8, marker='o')
plt.xlabel("K")
plt.ylabel("Inertia ")
Text(0, 0.5, 'Inertia ')
output_18_1.png
Kmeans = KMeans(n_clusters=5, init='k-means++', max_iter=300, n_init=10,random_state=0)
labels = Kmeans.fit_predict(X1_Matrix)
centroids1 = Kmeans.cluster_centers_
# the centroid points in each cluster
# Visualizing the 5 clusters
plt.scatter(x=X1_Matrix[labels==0, 0], y=X1_Matrix[labels==0, 1], s=20, c='red', marker='o')
plt.scatter(x=X1_Matrix[labels==1, 0], y=X1_Matrix[labels==1, 1], s=20, c='blue', marker='^')
plt.scatter(x=X1_Matrix[labels==2, 0], y=X1_Matrix[labels==2, 1], s=20, c='grey', marker='s')
plt.scatter(x=X1_Matrix[labels==3, 0], y=X1_Matrix[labels==3, 1], s=20, c='orange', marker='p')
plt.scatter(x=X1_Matrix[labels==4, 0], y=X1_Matrix[labels==4, 1], s=20, c='green', marker='*')
#Visualizing every centroids in different cluster.
plt.scatter(x=centroids1[:,0], y=centroids1[:,1], s=300, alpha=0.8, marker='+', label='Centroids')
#Style Setting
plt.title("Cluster Of Customers", fontsize=20)
plt.xlabel("Age")
plt.ylabel("Spending Score (1-100)")
plt.legend(loc=0)
output_19_1.png
pd.Series(labels).value_counts()
0 57
1 41
2 37
3 34
4 31
dtype: int64
inertias_2 = []
for i in range(1,8):
kmeans = KMeans(n_clusters=i, init='k-means++', max_iter=300, n_init=10,random_state=1)
kmeans.fit(X2_Matrix)
inertia = kmeans.inertia_
inertias_2.append(inertia)
print('For n_cluster =', i, 'The inertia is:', inertia)
For n_cluster = 1 The inertia is: 269981.28
For n_cluster = 2 The inertia is: 181363.59595959596
For n_cluster = 3 The inertia is: 106348.37306211118
For n_cluster = 4 The inertia is: 73679.78903948834
For n_cluster = 5 The inertia is: 44448.45544793371
For n_cluster = 6 The inertia is: 37233.81451071001
For n_cluster = 7 The inertia is: 30227.606513152015
# Creating the figure
figure = plt.figure(1, figsize=(15,6), dpi=80)
plt.plot(np.arange(1,8), inertias_2, alpha=0.8, marker='o')
plt.xlabel("K")
plt.ylabel("Inertia ")
Kmeans = KMeans(n_clusters=5, init='k-means++', max_iter=300, n_init=10,random_state=1)
labels = Kmeans.fit_predict(X2_Matrix)
centroids2 = Kmeans.cluster_centers_
output_22_0.png
# the centroid points in each cluster
# Visualizing the 5 clusters
plt.scatter(x=X2_Matrix[labels==0, 0], y=X1_Matrix[labels==0, 1], s=20, c='red', marker='o')
plt.scatter(x=X2_Matrix[labels==1, 0], y=X1_Matrix[labels==1, 1], s=20, c='blue', marker='^')
plt.scatter(x=X2_Matrix[labels==2, 0], y=X1_Matrix[labels==2, 1], s=20, c='grey', marker='s')
plt.scatter(x=X2_Matrix[labels==3, 0], y=X1_Matrix[labels==3, 1], s=20, c='orange', marker='p')
plt.scatter(x=X2_Matrix[labels==4, 0], y=X1_Matrix[labels==4, 1], s=20, c='green', marker='*')
#Visualizing every centroids in different cluster.
plt.scatter(x=centroids2[:,0], y=centroids2[:,1], s=300, alpha=0.8, marker='+', label='Centroids')
#Style Setting
plt.title("Cluster Of Customers", fontsize=20)
plt.xlabel("Annual Income (k$)")
plt.ylabel("Spending Score (1-100)")
plt.legend(loc=7)
output_23_1.png
5.总结
聚类结果显示：
在年龄方面，我们可以将客户分为5类，其中一类年轻人消费能力特别强，需要重点关注。
在年收入方面，我们可以将客户分为5类，有高收入低消费、高收入消费、中等收入中端消费、低收入第消费以及低收入高消费，可以针对他们做有针对性的营销策略。