K均值聚类,根据最接近的质心改变数据颜色

时间:2018-05-02 15:33:38

标签: python k-means

下面我的python代码对一组数据执行非常简单的K-means聚类。存在的问题是我需要根据最接近的质心来改变数据点的颜色。任何人都可以帮助我做我应该做的事情吗?

import numpy as np
import pandas as pd
from matplotlib import pyplot as plt


def calc_dist_euclidean(vec_1, vec_2):

    distances = np.sqrt(((vec_1 - vec_2[:, np.newaxis]) ** 2).sum(axis=2))
    dist_euclidean = np.argmin(distances, axis=0)

    return dist_euclidean


def init_cent(dataset, k):

    centroids = dataset.copy()
    np.random.shuffle(centroids)

    return centroids[:k]


def k_means(dataset, k):

    centroids = init_cent(dataset, k)

    distances = calc_dist_euclidean(dataset, centroids)

    cluster_assigning = np.array([dataset[distances == k].mean(axis=0) for k in range(centroids.shape[0])])

    return centroids, cluster_assigning


df = pd.read_csv('bristol_vacation_rentals_2016.csv')

dataset = df[['latitude', 'longitude']].values

k = 3

centroids, cluster_assigning = k_means(dataset, k)
plt.subplot(121)
plt.scatter(dataset[:, 0], dataset[:, 1], s=2)
plt.scatter(centroids[:, 0], centroids[:, 1],  c='r', s=50)
plt.xlabel('latitude')
plt.ylabel('longitude')

plt.subplot(122)
plt.scatter(dataset[:, 0], dataset[:, 1], s=2)
plt.scatter(cluster_assigning[:, 0], cluster_assigning[:, 1], c='r', s=50)
plt.xlabel('latitude')
plt.ylabel('longitude')
plt.show()

1 个答案:

答案 0 :(得分:0)

为了使其更容易,我将重用k_means()函数中的变量def k_means(dataset, k): centroids = init_cent(dataset, k) distances = calc_dist_euclidean(dataset, centroids) cluster_assigning = np.array([dataset[distances == k].mean(axis=0) for k in range(centroids.shape[0])]) return centroids, distances, cluster_assigning ,并将其返回:

colors

然后,您可以根据定义包含每种颜色的numpy数组colors[distances]的相应群集绘制质心和数据点,并使用df = pd.read_csv('bristol_vacation_rentals_2016.csv') dataset = df[['latitude', 'longitude']].values k = 3 colors = np.array(['r', 'g', 'b']) centroids, distances, cluster_assigning = k_means(dataset, k) plt.subplot(121) plt.scatter(dataset[:, 0], dataset[:, 1], c=colors[distances], s=2) plt.scatter(centroids[:, 0], centroids[:, 1], c=colors, s=50) plt.xlabel('latitude') plt.ylabel('longitude') plt.subplot(122) plt.scatter(dataset[:, 0], dataset[:, 1], c=colors[distances], s=2) plt.scatter(cluster_assigning[:, 0], cluster_assigning[:, 1], c=colors, s=50) plt.xlabel('latitude') plt.ylabel('longitude') plt.show() 为每个数据点创建一个numpy颜色数组:< / p>

export PATH=/usr/local/cuda-9.0/bin${PATH:+:${PATH}}
        export LD_LIBRARY_PATH=/usr/local/cuda-9.0/lib64\
                   ${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}