kemans聚类实战kemans聚类实战

#引入依赖 import numpy as np import pandas as pd import matplotlib.pyplot as plt import plotly.offline as py import plotly.graph_objects as go from sklearn import preprocessing from sklearn.cluster import KMeans df = pd.read_csv(r'data.csv') print(df.head(5)) print(df.info()) trace = go.Scatter(x=df['Annual Income (k$)'], y=df['Spending Score (1-100)'], mode='markers') layout = go.Layout(xaxis=dict(title='Annual Income (k$)'), yaxis=dict(title='Spending Score (1-100)')) plt = go.Figure(data=[trace], layout=layout) # fig.show() plt.show() # 求出簇内误方差 data = df[['Annual Income (k$)', 'Spending Score (1-100)']] sse = [] k_range = range(1, 10) for k in k_range: km = KMeans(n_clusters=k) km.fit(data) sse.append(km.inertia_) print(sse) # 利用“肘”方法确定最佳k值 trace = go.Scatter(x=list(k_range), y=sse) layout = go.Layout(xaxis=dict(title='k'), yaxis=dict(title='sse')) plt = go.Figure(data=[trace], layout=layout) plt.show() # 划分客户类型 km = KMeans(n_clusters = 5) predicted = km.fit_predict(data) df['cluster'] = predicted print(df.head(5)) df0 = df[df['cluster'] == 0] df1 = df[df['cluster'] == 1] df2 = df[df['cluster'] == 2] df3 = df[df['cluster'] == 3] df4 = df[df['cluster'] == 4] trace0 = go.Scatter(x=df0['Annual Income (k$)'], y=df0['Spending Score (1-100)'], mode='markers', marker=dict(color='purple'), name='中等收入-中等消费') trace1 = go.Scatter(x=df1['Annual Income (k$)'], y=df1['Spending Score (1-100)'], mode='markers', marker=dict(color='orange'), name='低收入-低消费') trace2 = go.Scatter(x=df2['Annual Income (k$)'], y=df2['Spending Score (1-100)'], mode='markers', marker=dict(color='yellow'), name='低收入-高消费') trace3 = go.Scatter(x=df3['Annual Income (k$)'], y=df3['Spending Score (1-100)'], mode='markers', marker=dict(color='green'), name='高收入-低消费') trace4 = go.Scatter(x=df4['Annual Income (k$)'], y=df4['Spending Score (1-100)'], mode='markers', marker=dict(color='blue'), name='高收入-高消费') fig = go.Figure() fig.add_trace(trace0) fig.add_trace(trace1) fig.add_trace(trace2) fig.add_trace(trace3) fig.add_trace(trace4) fig.update_layout(xaxis=dict(title='Annual Income (k$)'), yaxis=dict(title='Spending Score (1-100)')) fig.show() # 簇内质心 df_center = km.cluster_centers_ df_center = pd.DataFrame(df_center, columns=['x','y']) print("簇内质心",df_center) fig.add_trace(go.Scatter(x=df_center['x'], y=df_center['y'], mode='markers', marker=dict(color='red', line=dict(width=1)), name='不同簇的质心')) fig.show() # 四、结论¶ # 根据年收入与消费积分的关系，得出5种客户类型： # # 客户群0 中等收入-中等消费 一般价值客户 # 客户群1 低收入-低消费 低价值客户 # 客户群2 低收入-高消费 重要保持客户 # 客户群3 高收入-低消费 重要发展客户 # 客户群4 高收入-高消费 重要价值客户