kmeans聚类算法python实现

好的，下面给出 kmeans 聚类算法的 Python 实现代码：

import numpy as np
import matplotlib.pyplot as plt

class KMeans:
    def __init__(self, k=2, max_iter=100):
        self.k = k                  # 聚类数
        self.max_iter = max_iter    # 最大迭代次数

    def fit(self, X):
        self.X = X
        self.centroids = self.initialize_centroids()  # 初始化聚类中心
        for i in range(self.max_iter):
            self.clusters = self.create_clusters()    # 创建聚类
            prev_centroids = self.centroids
            self.update_centroids()                   # 更新聚类中心
            if self.is_converged(prev_centroids):      # 判断是否收敛
                break

    def initialize_centroids(self):
        centroids = self.X[np.random.choice(range(len(self.X)), self.k, replace=False)]
        return centroids

    def create_clusters(self):
        clusters = [[] for _ in range(self.k)]
        for i, x in enumerate(self.X):
            distances = [np.linalg.norm(x - c) for c in self.centroids]
            cluster_idx = np.argmin(distances)
            clusters[cluster_idx].append(i)
        return clusters

    def update_centroids(self):
        for i, cluster in enumerate(self.clusters):
            self.centroids[i] = np.mean(self.X[cluster], axis=0)

    def is_converged(self, prev_centroids):
        distances = [np.linalg.norm(self.centroids[i] - prev_centroids[i]) for i in range(self.k)]
        return sum(distances) == 0

    def predict(self, X):
        distances = [np.linalg.norm(X - c) for c in self.centroids]
        cluster_idx = np.argmin(distances)
        return cluster_idx

if __name__ == '__main__':
    # 生成数据
    np.random.seed(0)
    X = np.random.randn(100, 2)
    X[:50, :] -= 2*np.ones((50, 2))
    X[50:, :] += 2*np.ones((50, 2))

    # 聚类
    kmeans = KMeans(k=2)
    kmeans.fit(X)

    # 可视化结果
    plt.scatter(X[:, 0], X[:, 1], c=[kmeans.predict(x) for x in X])
    plt.scatter(kmeans.centroids[:, 0], kmeans.centroids[:, 1], marker='x', s=200, linewidths=3, color='r')
    plt.show()

这是一个简单的 kmeans 聚类算法实现，你可以根据自己的需求进行修改。