Kmeans python 实现

# -*- coding: utf-8 -*-

"""

Created on Tue Jul 26 15:16:42 2016

@author: brian

"""

import numpy as np

import matplotlib.pyplot as plt

def loadData(file = "testSet.txt"):

dataMat = np.loadtxt(file)

return dataMat

def showData(data):

plt.plot(data[:,0],data[:,1],"bo")

def classify(dataSet , centers):

"""

dataSet Numbers * dims; centers: numbers * dims

"""

k = np.shape(centers)[0] # k 类

n = np.shape(dataSet)[0] # 样本数

dim =np.shape(dataSet)[1]# 样本维度

diss = np.zeros((n,k))

for i in range(k):

dis_row = dataSet - centers[i,:]

seq = dis_row**2

dis = (seq[:,0] + seq[:,1])**(0.5)

diss[:,i] = dis

return np.argmin(diss,axis = 1)

def randCent(dataSet, k):

"""

随机选取Ｋ个点，　保证Ｋ处于样本数据之间。　

"""

n = np.shape(dataSet)[1]

centroids = np.mat(np.zeros((k,n)))#create centroid mat

for j in range(n):

#create random cluster centers, within bounds of each dimension

minJ = min(dataSet[:,j])

rangeJ = float(max(dataSet[:,j]) - minJ)

centroids[:,j] = np.mat(minJ + rangeJ * np.random.rand(k,1))

return np.asarray(centroids)

def updataCents(datas):

return datas.mean(axis = 0)

def Kmeans(data,k=4,iteration = 10):

cents = randCent(data, k)

# in iterator for j in range(10):

for j in range(3):

clasIndex = classify(data, cents)

plt.figure(j)

for i in range(k):

if i == 0:

data_i = data[np.where(clasIndex==i)]

plt.plot(data_i[:,0],data_i[:,1],"bo")

plt.plot(cents[i,0],cents[i,1],"bx" , markersize=12 ,markeredgewidth =3)

elif i == 1:

data_i = data[np.where(clasIndex==i)]

plt.plot(data_i[:,0],data_i[:,1],"go")

plt.plot(cents[i,0],cents[i,1],"gx" ,markersize=12 , markeredgewidth=3)

elif i == 2:

data_i = data[np.where(clasIndex==i)]

plt.plot(data_i[:,0],data_i[:,1],"ro")

plt.plot(cents[i,0],cents[i,1],"rx" , markersize=12 ,markeredgewidth =3)

elif i == 3:

data_i = data[np.where(clasIndex==i)]

plt.plot(data_i[:,0],data_i[:,1],"co")

plt.plot(cents[i,0],cents[i,1],"cx" , markersize=12 ,markeredgewidth =3)

cents[i,:] = updataCents(data[np.where(clasIndex==i)])

plt.show()

k=4

data = loadData()

Kmeans(data)