亲宝软件园·资讯

展开

python实现k-means算法

咕嘟咕嘟_ 人气:0

聚类属于无监督学习,K-means算法是很典型的基于距离的聚类算法,采用距离作为相似性的评价指标,即认为两个对象的距离越近,其相似度就越大。该算法认为簇是由距离靠近的对象组成的,因此把得到紧凑且独立的簇作为最终目标。

下面来看看python实现k-means算法的详细代码吧:

# -*- coding:utf-8 -*-
import random

import numpy as np
from matplotlib import pyplot


class K_Means(object):
    # k是分组数;tolerance‘中心点误差';max_iter是迭代次数
    def __init__(self, k=2, tolerance=0.0001, max_iter=300):
        self.k_ = k
        self.tolerance_ = tolerance
        self.max_iter_ = max_iter

    def fit(self, data):
        self.centers_ = {}
        for i in range(self.k_):
            self.centers_[i] = data[random.randint(0,len(data))]
        # print('center', self.centers_)
        for i in range(self.max_iter_):
            self.clf_ = {} #用于装归属到每个类中的点[k,len(data)]
            for i in range(self.k_):
                self.clf_[i] = []
            # print("质点:",self.centers_)
            for feature in data:
                distances = [] #装中心点到每个点的距离[k]
                for center in self.centers_:
                    # 欧拉距离
                    distances.append(np.linalg.norm(feature - self.centers_[center]))
                classification = distances.index(min(distances))
                self.clf_[classification].append(feature)

            # print("分组情况:",self.clf_)
            prev_centers = dict(self.centers_)

            for c in self.clf_:
                self.centers_[c] = np.average(self.clf_[c], axis=0)

            # '中心点'是否在误差范围
            optimized = True
            for center in self.centers_:
                org_centers = prev_centers[center]
                cur_centers = self.centers_[center]
                if np.sum((cur_centers - org_centers) / org_centers * 100.0) > self.tolerance_:
                    optimized = False
            if optimized:
                break

    def predict(self, p_data):
        distances = [np.linalg.norm(p_data - self.centers_[center]) for center in self.centers_]
        index = distances.index(min(distances))
        return index


if __name__ == '__main__':
    x = np.array([[1, 2], [1.5, 1.8], [5, 8], [8, 8], [1, 0.6], [9, 11]])
    k_means = K_Means(k=2)
    k_means.fit(x)
    for center in k_means.centers_:
        pyplot.scatter(k_means.centers_[center][0], k_means.centers_[center][1], marker='*', s=150)

    for cat in k_means.clf_:
        for point in k_means.clf_[cat]:
            pyplot.scatter(point[0], point[1], c=('r' if cat == 0 else 'b'))

    predict = [[2, 1], [6, 9]]
    for feature in predict:
        cat = k_means.predict(feature)
        pyplot.scatter(feature[0], feature[1], c=('r' if cat == 0 else 'b'), marker='x')

    pyplot.show()

加载全部内容

相关教程
猜你喜欢
用户评论