Kmeans.py

from numpy import *
import matplotlib.pyplot as plt


def loadDataSet(fileName):
    dataSet = []
    f = open(fileName)
    for line in f.readlines():
        curLine = line.strip().split(',')  # 这里","表示以文件中数据之间的分隔符","分割字符串
        row = []
        for item in curLine:
            row.append(float(item))
        dataSet.append(row)

    return mat(dataSet)


# 求向量距离
def distEclud(vecA, vecB):
    return sqrt(sum(power(vecA - vecB, 2)))


# 随机生成k个点作为初始质心
# 若选择随机生成的点作为初始质心，则有可能导致后面更新质心时出现有的质心为 NaN(非数) 的情况
# def initCent(dataSet, k):
#     n = shape(dataSet)[1]  # n是列数
#     centroids = mat(zeros((k, n)))
#     for j in range(n):
#         minJ = min(dataSet[:, j])  # 找到第j列最小值
#         rangeJ = max(dataSet[:, j]) - minJ  # 求第j列最大值与最小值的差
#         centroids[:, j] = minJ + random.rand(k, 1) * rangeJ  # 生成k行1列的在(0, 1)之间的随机数矩阵
#     return centroids

# 选前k个点作为初始质心
def initCent(dataSet, k):
    data = []
    for i in range(k):
        data.append(dataSet[i].tolist())
    a = array(data)
    centroids = mat(a)
    return centroids


# K均值聚类算法实现
def KMeans(dataSet, k, distMeas=distEclud):
    m = shape(dataSet)[0]  # 数据集的行
    clusterAssment = mat(zeros((m, 2)))
    centroids = initCent(dataSet, k)
    clusterChanged = True
    while clusterChanged:
        clusterChanged = False
        for i in range(m):  # 遍历数据集中的每一行数据
            minDist = inf
            minIndex = -1
            for j in range(k):  # 寻找最近质心
                distJI = distMeas(centroids[j, :], dataSet[i, :])
                if distJI < minDist:  # 更新最小距离和质心下标
                    minDist = distJI
                    minIndex = j
            if clusterAssment[i, 0] != minIndex:
                clusterChanged = True
            clusterAssment[i, :] = minIndex, minDist ** 2  # 记录最小距离质心下标，最小距离的平方
        for cent in range(k):  # 更新质心位置
            ptsInClust = dataSet[nonzero(clusterAssment[:, 0].A == cent)[0]]  # 获得距离同一个质心最近的所有点的下标，即同一簇的坐标
            centroids[cent, :] = mean(ptsInClust, axis=0)  # 求同一簇的坐标平均值，axis=0表示按列求均值

    return centroids, clusterAssment


# 取数据的前两维特征作为该条数据的x , y 坐标，
def getXY(dataSet):
    import numpy as np
    m = shape(dataSet)[0]  # 数据集的行
    X = []
    Y = []
    for i in range(m):
        X.append(dataSet[i, 0])
        Y.append(dataSet[i, 1])
    return np.array(X), np.array(Y)


# 数据可视化
def showCluster(dataSet, k, clusterAssment, centroids):
    fig = plt.figure()
    plt.title("K-means")
    ax = fig.add_subplot(111)
    data = []

    for cent in range(k):  # 提取出每个簇的数据
        ptsInClust = dataSet[nonzero(clusterAssment[:, 0].A == cent)[0]]  # 获得属于cent簇的数据
        data.append(ptsInClust)

    for cent, c, marker in zip(range(k), ['r', 'g', 'b', 'y'], ['^', 'o', '*', 's']):  # 画出数据点散点图
        X, Y = getXY(data[cent])
        ax.scatter(X, Y, s=80, c=c, marker=marker)

    centroidsX, centroidsY = getXY(centroids)
    ax.scatter(centroidsX, centroidsY, s=1000, c='black', marker='+', alpha=1)  # 画出质心点
    ax.set_xlabel('X Label')
    ax.set_ylabel('Y Label')
    plt.show()


if __name__ == "__main__":
    cluster_Num = 3
    data = loadDataSet("test_kmeans.txt")
    print(data)
    centroids, clusterAssment = KMeans(data, cluster_Num)