kmeans聚類簇個數選擇

藉助sklearn庫實現kmeans聚類和輪廓係數計算

from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_score
K = range(2,20) # 設置主題個數區間
coef = [] 
for k in K:
    km = KMeans(n_clusters=k,random_state = 0).fit(feature) #構建kmeans模型並訓練
    score = silhouette_score(feature, km.labels_,sample_size=1000) # 計算對應模型的輪廓係數
    coef.append(score)

可視化：

import matplotlib.pyplot as plt
%matplotlib inline
plt.plot(K,coef) # K爲x軸輸出，coef是y軸輸出
plt.show()