sklearn

# 聚类前
X = np.random.rand(100, 2)
plt.scatter(X[:, 0], X[:, 1], marker='o')
plt.show()

# 聚类后
kmeans = KMeans(n_clusters=2).fit(X)
label_pred = kmeans.labels_
plt.scatter(X[:, 0], X[:, 1], c=label_pred)
plt.show()

k均值聚类

# -*- coding: UTF-8 -*-import matplotlib.pyplot as plt
import numpy as npdef distEclud(vecA, vecB):'''欧氏距离计算函数:param vecA::param vecB::return: float'''dist = 0.0# ========= show me your code ==================# heredist = np.sqrt(np.sum(vecA - vecB)**2 )# ========= show me your code ==================return distdef randCent(dataMat, k):'''为给定数据集构建一个包含K个随机质心的集合,随机质心必须要在整个数据集的边界之内,这可以通过找到数据集每一维的最小和最大值来完成然后生成0到1.0之间的随机数并通过取值范围和最小值,以便确保随机点在数据的边界之内:param np.dataMat::param k::return: np.dataMat'''# 获取样本数与特征值m, n = np.shape(dataMat)# 初始化质心,创建(k,n)个以零填充的矩阵centroids = np.mat(np.zeros((k, n)))print(centroids)# ========= show me your code ==================# 循环遍历特征值# herefor i in range(n - 1):mindata = dataMat[:, i].min()maxdata = dataMat[:, i].max()diffdata = maxdata - mindatacentroids[i] = mindata + np.random.random(k) * diffdata# ========= show me your code ==================# 返回质心return centroids.Adef kMeans(dataMat, k, distMeas=distEclud):'''创建K个质心,然后将每个点分配到最近的质心,再重新计算质心。这个过程重复数次,直到数据点的簇分配结果不再改变为止:param dataMat: 数据集:param k: 簇的数目:param distMeans: 计算距离:return:'''# 获取样本数和特征数m, n = np.shape(dataMat)# 初始化一个矩阵来存储每个点的簇分配结果# clusterAssment包含两个列:一列记录簇索引值,第二列存储误差(误差是指当前点到簇质心的距离,后面会使用该误差来评价聚类的效果)clusterAssment = np.mat(np.zeros((m, 2)))# 创建质心,随机K个质心centroids = randCent(dataMat, k)# 初始化标志变量,用于判断迭代是否继续,如果True,则继续迭代clusterChanged = Truewhile clusterChanged:clusterChanged = False# 遍历所有数据找到距离每个点最近的质心,# 可以通过对每个点遍历所有质心并计算点到每个质心的距离来完成for i in range(m):minDist = float("inf")minIndex = -1for j in range(k):# 计算数据点到质心的距离# 计算距离是使用distMeas参数给出的距离公式,默认距离函数是distEcluddistJI = distMeas(centroids[j, :], dataMat[i, :])# 如果距离比minDist(最小距离)还小,更新minDist(最小距离)和最小质心的index(索引)if distJI < minDist:minDist = distJIminIndex = j# 如果任一点的簇分配结果发生改变,则更新clusterChanged标志# ========= show me your code ==================# hereif clusterAssment[i, 0] != minIndex:clusterChanged = True# ========= show me your code ==================# 更新簇分配结果为最小质心的index(索引),minDist(最小距离)的平方clusterAssment[i, :] = minIndex, minDist ** 2print(centroids)# 遍历所有质心并更新它们的取值# ========= show me your code ==================# herefor j in range(k):pointsInCluster = dataMat[np.nonzero(clusterAssment[:, 0].A == j)[0]]  # 获取簇类所有的点centroids[j, :] = np.mean(pointsInCluster, axis=0)   # 对矩阵的行求均值# ========= show me your code ==================# 返回所有的类质心与点分配结果return centroids, clusterAssment
if __name__ == '__main__':X = np.random.rand(100, 2)
# 运行Kmeans，假设有两聚类中心center,label_pred = kMeans(X, k=2)
# 将标签转化成易绘图的形式label = label_pred[:, 0].A.reshape(-1)
# 将结果可视化plt.scatter(X[:, 0], X[:, 1], c=label)plt.scatter(center[0, 0], center[0, 1], marker="*", s=100)plt.scatter(center[1, 0], center[1, 1], marker="*", s=100)plt.show()

参考文献： 西瓜书 维基百科 机器学习实战https://www.zybuluo.com/rianusr/note/1199877http://ddrv.cn/a/66611https://blog.csdn.net/zhouxianen1987/article/details/68945844http://ddrv.cn/a/66611https://zhuanlan.zhihu.com/p/29538307