階層式分群
1凫海、使用scipy繪制樹狀圖
from sklearn.datasets import load_iris
iris=load_iris()
import scipy.cluster.hierarchy as sch
import matplotlib.pyplot as plt
dendrogram = sch.dendrogram(sch.linkage(iris.data,method='ward'))
# method='ward' 其他選擇看sch.linkage幫助家乘，常用ward
plt.title('Dendrogram')
plt.xlabel('Iris')
plt.ylabel('Euclidean distances')
plt.show()
樹狀圖
2科汗、使用sklearn分群
from sklearn.cluster import AgglomerativeClustering
hc = AgglomerativeClustering(n_clusters=3,affinity='euclidean',linkage='ward')
# n_clusters 一般不設(shè)置卷玉，linkage常用ward丈牢，具體看函數(shù)幫助
y_hc = hc.fit_predict(iris.data)
plt.scatter(iris.data[y_hc==0,2],iris.data[y_hc==0,3],s=100,c='red',label='Cluster1')
plt.scatter(iris.data[y_hc==1,2],iris.data[y_hc==1,3],s=100,c='green',label='Cluster2')
plt.scatter(iris.data[y_hc==2,2],iris.data[y_hc==2,3],s=100,c='blue',label='Cluster3')
# 選擇y_hc 分別為0鳞青，1衙四，2的分組铃肯，使用iris.data的第3列（petal.length)，第4列(petal.width)數(shù)據(jù)畫圖
plt.title('Culsters of Iris')
plt.xlabel('Petal.Length')
plt.ylabel('Petal.Width')
plt.legend()
plt.show()
sklearn分群
k-means 分群
from sklearn.datasets import load_iris
iris=load_iris()
from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters = 3,init = 'k-means++',random_state=123)
# n_clusters指定分為幾個(gè)群传蹈，其他參數(shù)設(shè)置查看KMeans幫助押逼，
y_kmeans = kmeans.fit_predict(iris.data)
import matplotlib.pyplot as plt
plt.scatter(iris.data[y_kmeans==0,2],iris.data[y_kmeans==0,3],s=100,c='red',label='Cluster1')
plt.scatter(iris.data[y_kmeans==1,2],iris.data[y_kmeans==1,3],s=100,c='green',label='Cluster2')
plt.scatter(iris.data[y_kmeans==2,2],iris.data[y_kmeans==2,3],s=100,c='blue',label='Cluster3')
plt.scatter(kmeans.cluster_centers_[:,2],kmeans.cluster_centers_[:,3],s=100,c='yellow',label='center')
# kmeans.cluster_centers_ 得到3個(gè)群的中心點(diǎn)
plt.title('Culsters of Iris')
plt.xlabel('Petal.Length')
plt.ylabel('Petal.Width')
plt.legend()
plt.show()
Kmeans分群
DBSCAN 分群
from sklearn.datasets import load_iris
iris=load_iris()
from sklearn.cluster import DBSCAN
dbs = DBSCAN(eps=1,min_samples=3)
# 無法設(shè)置分為幾個(gè)群，eps點(diǎn)范圍半徑惦界，半徑內(nèi)點(diǎn)成群挑格，min_samples 最小3個(gè)成群
y_dbs = dbs.fit_predict(iris.data)
import matplotlib.pyplot as plt
plt.scatter(iris.data[y_dbs==0,2],iris.data[y_dbs==0,3],s=100,c='red',label='Cluster1')
plt.scatter(iris.data[y_dbs==1,2],iris.data[y_dbs==1,3],s=100,c='green',label='Cluster2')
plt.title('Culsters of Iris')
plt.xlabel('Petal.Length')
plt.ylabel('Petal.Width')
plt.legend()
plt.show()
DBSCAN分群
k-means 和DBSCAN 對(duì)比
import numpy as np
from PIL import Image
img=Image.open('handwrite.png').convert('L').rotate(-90)
imgarr = np.array(img)
# convert('L') 將圖片信息由三維轉(zhuǎn)成二維，rotate(-90)將圖片逆時(shí)針轉(zhuǎn)90度
from sklearn.preprocessing import binarize
imagedata = np.where(binarize(imgarr,0)==0) # 將圖片信息轉(zhuǎn)成布爾型
import matplotlib.pyplot as plt
plt.scatter(imagedata[0],imagedata[1],s=100,c='red',label='Cluster1')
plt.legend()
plt.show()
將1和8 分開
使用kmeans
from sklearn.cluster import KMeans
X = np.column_stack([imagedata[0],imagedata[1]])
kmeans = KMeans(n_clusters=2,init='k-means++',random_state=137)
y_kmeans = kmeans.fit_predict(X)
plt.scatter(X[y_kmeans==0,0],X[y_kmeans==0,1],s=100,c='red',label='Cluster1')
plt.scatter(X[y_kmeans==1,0],X[y_kmeans==1,1],s=100,c='blue',label='Cluster2')
plt.scatter(kmeans.cluster_centers_[:,0],kmeans.cluster_centers_[:,1],s=100,c='yellow',label='central')
plt.legend()
plt.show()
kmeans分群
使用DBSCAN
from sklearn.cluster import DBSCAN
dbs =DBSCAN(eps=1,min_samples =3)
y_dbs = dbs.fit_predict(X)
plt.scatter(X[y_dbs==0,0],X[y_dbs==0,1],s=100,c='red',label='Cluster1')
plt.scatter(X[y_dbs==1,0],X[y_dbs==1,1],s=100,c='blue',label='Cluster2')
plt.legend()
plt.show()
DBSCAN分群
分群結(jié)果評(píng)價(jià)
階層式分群和kmeans分群都要事先輸入分成幾個(gè)群沾歪，以下方法主要判斷分為幾個(gè)群比較好漂彤。
方法一 WCSS（with-in cluster sum of square)
import pandas
dataset = pandas.read_csv('Data/customers.csv')
dataset['sex'] = dataset['Genre'].map(lambda e:1 if e=='Male' else 0) # 數(shù)據(jù)預(yù)處理
del dataset['Genre'],dataset['CustomerID'] # 數(shù)據(jù)預(yù)處理
X = dataset.values
from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters = 5,init='k-means++',random_state=42)
y_kmeans = kmeans.fit_predict(X)
print(kmeans.inertia_) # kmeans.inertia_ 就是這次kmeans模型的WCSS
import matplotlib.pyplot as plt
wcss= []
for i in range(1,11):
kmeans = KMeans(n_clusters= i ,init='k-means++',random_state=42) #將數(shù)據(jù)分成1-10個(gè)群，分別計(jì)算WCSS灾搏，然后畫圖
kmeans.fit(X)
wcss.append(kmeans.inertia_)
plt.plot(range(1,11),wcss)
plt.title('The Elbow Method')
plt.xlabel('Number of clusters')
plt.ylabel('WCSS')
plt.show()
得到的WCSS圖挫望，拐點(diǎn)處為最佳的分群數(shù)量（很難看出來，個(gè)人覺得這個(gè)方法沒什么用）
WCSS
方法二 silhouette coefficient
silhouette值越高确镊，分群效果越好
from sklearn import metrics
import matplotlib.pyplot as plt
sil = []
for i in range(2,11):
kmeans= KMeans(n_clusters=i,init='k-means++',random_state=42) # 分成2-10群士骤，分別計(jì)算silhouette值，畫圖
y_kmeans = kmeans.fit_predict(X)
sil.append(metrics.silhouette_score(X,y_kmeans))
plt.plot(range(2,11),sil)
plt.title('The Silhouette Method')
plt.xlabel('Number of clusters')
plt.ylabel('Silhouette Coefficient')
plt.show()
silhouette method
通過silhouette值比較不同分群方法
from sklearn.cluster import AgglomerativeClustering
from sklearn.cluster import KMeans
ward = AgglomerativeClustering(n_clusters=5,affinity='euclidean',linkage='ward')
y_ward = ward.fit_predict(X)
complete = AgglomerativeClustering(n_clusters=5,affinity='euclidean',linkage='complete')
y_complete = complete.fit_predict(X)
kmeans = KMeans(n_clusters=5,init='k-means++',random_state=42)
y_kmeans= kmeans.fit_predict(X)
for est,title in zip([y_ward,y_complete,y_kmeans],['ward','complete','kmeans']):
print(title,metrics.silhouette_score(X,est))
# ward 0.439975272125
# complete 0.439975272125
# kmeans 0.444242912753
# kmeans方法的silhouette值最高蕾域，分群效果最好
