如何在k-means中记录每一次迭代的质心？

import numpy as np

def k_means(data, k, threshold):
    # 初始化k个质心
    centroids = np.random.choice(data, size=k, replace=False)
    centroids_history = [centroids]  # 记录质心的历史

    while True:
        # 分配数据点到最近的质心簇
        clusters = [[] for _ in range(k)]
        for point in data:
            distances = [np.linalg.norm(point - centroid) for centroid in centroids]
            cluster_index = np.argmin(distances)
            clusters[cluster_index].append(point)

        # 计算新的质心
        new_centroids = []
        for cluster in clusters:
            new_centroids.append(np.mean(cluster, axis=0))

        # 检查质心的变化
        if np.linalg.norm(np.array(new_centroids) - np.array(centroids)) < threshold:
            break

        centroids = new_centroids
        centroids_history.append(centroids)

    return centroids_history

# 示例使用
data = np.array([[1, 2], [2, 1], [10, 12], [12, 10], [20, 25], [25, 20]])
k = 2
threshold = 0.01

centroids_history = k_means(data, k, threshold)
for i, centroids in enumerate(centroids_history):
    print(f"Iteration {i+1}: {centroids}")