Python实现的Kmeans++算法实例

    cx = (max_x + min_x) / 2
    cy = (max_y + min_y) / 2

    print "%%!PS-Adobe-3.0n%%%%BoundingBox: -5 -5 %d %d" % (W + 10, H + 10)

    print ("/l {rlineto} def /m {rmoveto} defn" +
           "/c { .25 sub exch .25 sub exch .5 0 360 arc fill } defn" +
           "/s { moveto -2 0 m 2 2 l 2 -2 l -2 -2 l closepath " +
           "   gsave 1 setgray fill grestore gsave 3 setlinewidth" +
           " 1 setgray stroke grestore 0 setgray stroke }def")

    for i, cc in enumerate(cluster_centers):
        print ("%g %g %g setrgbcolor" %
               (colors[i].r, colors[i].g, colors[i].b))

        for p in points:
            if p.group != i:
                continue
            print ("%.3f %.3f c" % ((p.x - cx) * scale + W / 2,
                                    (p.y - cy) * scale + H / 2))

        print ("n0 setgray %g %g s" % ((cc.x - cx) * scale + W / 2,
                                        (cc.y - cy) * scale + H / 2))

    print "n%%%%EOF"

def main():
    npoints = 30000
    k = 7 # # clusters

    points = generate_points(npoints, 10)
    cluster_centers = lloyd(points, k)
    print_eps(points, cluster_centers)

main()

上述代码实现的算法是针对二维数据的，所以Point对象有三个属性，分别是在x轴上的值、在y轴上的值、以及所属的簇的标识。函数lloyd是kmeans++算法的整体实现，其先是通过kpp函数选取合适的种子点，然后对数据集实行kmeans算法进行聚类。kpp函数的实现完全符合上述kmeans++的基本思路的2、3、4步。

4、matlab版本的kmeans++

function [L,C] = kmeanspp(X,k)
%KMEANS Cluster multivariate data using the k-means++ algorithm.
%   [L,C] = kmeans_pp(X,k) produces a 1-by-size(X,2) vector L with one class
%   label per column in X and a size(X,1)-by-k matrix C containing the