机器学习_K-Means

Sklearn实现K-Means

import numpy as np
import matplotlib.pyplot as plt
import matplotlib as mpl
import scipy.io
from sklearn.cluster imoprt KMeans

导入数据

def loaddata():
	data = np.loadtxt('data/cluster_data.csv',delimiter=',')
	return data
X = loaddata()

fit模型

model = KMeans(n_cluster=3,max_iter=10)
model.fit(X)

print('聚类中心：',model.cluster_centers_)
print('每个样本所属的簇：', model.labesl_)

# 画图
cm_dark = mpl.colors.ListedColormap(['g', 'r', 'b'])
plt.scatter(X[:, 0], X[:, 1], c=model.labels_, cmap=cm_dark, s=20)
plt.scatter(model.cluster_centers[:, 0], model.cluster_centers_[:, 1], c=np.arange(len(model.cluster_centers_)))

层次聚类实现

from sklearn.cluster import AgglomerativeClustering

linkage可取值：

• ward：最小方差
• complete：最大距离
• average：平均距离
• single：最小距离

model = AgglomerativeClustering(n_clusters=3,affinity='euclidean',linkage='complete')
print('每个样本所属的簇：',model.labels_)

# 画图
cm_dark = mpl.colors.ListedColormap(['g', 'r', 'b'])
plt.scatter(X[:, 0], X[:, 1], c=model.labels_, cmap=cm_dark, s=20)
plt.show()

密度聚类

根据密度聚类的过程，无需设置聚类个数

from sklearn.cluster import DBSCAN

def loaddata():
	data = np.loadtxt('data/cluster_data.csv',delimiter=',')
	return data
X = loaddata()

model = DBSCAN(eps=0.5, min_samples=5, metric='euclidean')
model.fit(X)
print('每个样本所属的簇', model.labels_)

画图

cm_dark = mpl.colors.ListedColormap(['g', 'r', 'b', 'c'])
plt.scatter(X[:,0], X[:,1], c=model.labels_, cmap=cm_dark, s=20)
plt.show()