kmeans聚类算法python代码

下面是一个基本的k-means聚类算法的Python代码实现。假设我们有一个包含n个数据点的数据集，每个数据点有d个特征。我们将数据集分成k个簇，然后根据数据点的相似性将每个数据点分配到一个簇中。

import numpy as np

class KMeans:
    def __init__(self, k=2, tol=0.001, max_iter=300):
        self.k = k
        self.tol = tol
        self.max_iter = max_iter

    def fit(self, X):
        self.centroids = {}

        # Initialize centroids
        for i in range(self.k):
            self.centroids[i] = X[i]

        # Run k-means algorithm
        for i in range(self.max_iter):
            self.clusters = {}

            # Initialize clusters
            for j in range(self.k):
                self.clusters[j] = []

            # Assign points to clusters
            for x in X:
                distances = [np.linalg.norm(x - self.centroids[c]) for c in self.centroids]
                cluster = distances.index(min(distances))
                self.clusters[cluster].append(x)

            # Update centroids
            prev_centroids = dict(self.centroids)

            for cluster in self.clusters:
                self.centroids[cluster] = np.average(self.clusters[cluster], axis=0)

            # Check convergence
            is_converged = True

            for c in self.centroids:
                original_centroid = prev_centroids[c]
                current_centroid = self.centroids[c]

                if np.sum((current_centroid - original_centroid) / original_centroid * 100.0) > self.tol:
                    is_converged = False

            if is_converged:
                break

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

这个实现中，我们首先初始化k个簇的质心，然后根据每个数据点与质心的距离将它们分配到不同的簇中。接着，我们更新每个簇的质心，重新分配数据点，并检查算法是否收敛。最后，我们可以使用`predict()`方法将新的数据点分配到一个簇中。