sklearn 和从头开始的不同 Kmean 结果

Question

我尝试比较 sklearn 包和从头开始的 kmean 聚类结果。 scratch代码如下：

import matplotlib.pyplot as plt
from matplotlib import style

style.use('ggplot')
import numpy as np

colors = 10 * ["g", "r", "c", "b", "k"]


class K_Means:
    def __init__(self, k=3, tol=0.001, max_iter=300):
        self.k = k
        self.tol = tol
        self.max_iter = max_iter

    def fit(self, data):

        self.centroids = {}

        for i in range(self.k):
            self.centroids[i] = data[i]

        for i in range(self.max_iter):
            self.classifications = {}

            for i in range(self.k):
                self.classifications[i] = []

            for featureset in data:
                distances = [np.linalg.norm(featureset - self.centroids[centroid]) for centroid in self.centroids]
                classification = distances.index(min(distances))
                self.classifications[classification].append(featureset)

            prev_centroids = dict(self.centroids)

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

            optimized = 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:
                    print(np.sum((current_centroid - original_centroid) / original_centroid * 100.0))
                    optimized = False

            if optimized:
                break

    def predict(self, data):
        distances = [np.linalg.norm(data - self.centroids[centroid]) for centroid in self.centroids]
        classification = distances.index(min(distances))
        return classification

kmeans = K_Means()
kmeans.fit(reduced_data)

for centroid in kmeans.centroids:
    plt.scatter(kmeans.centroids[centroid][0], kmeans.centroids[centroid][1],
                marker="x", color="b", s=169, linewidths=3, zorder=10)

for classification in kmeans.classifications:
    color = colors[classification]
    for featureset in kmeans.classifications[classification]:
        plt.scatter(featureset[0], featureset[1], marker="o", color=color)

plt.show()

但是，由于收敛的质心不同，结果也不同。来自 sklearn 的散点图：

同时，上面代码的散点图：我想知道 scratch 代码中有什么错误。

Answer 1

K-means 高度依赖于初始化条件，即方法的起点。 scikit-learn可以根据数据进行智能初始化。如果您仔细阅读文档，您可能可以配置 scikit-learn 的版本以匹配您自己的版本。另外，请尝试查看 source code 以获取更多线索。

sklearn 和从头开始的不同 Kmean 结果

Different Kmean results by sklearn and from scratch

python

cluster-analysis

k-means

scikit-learn