Python中使用DBSCAN算法对坐标点进行聚类例子

在服务器安装了Anaconda4，Anaconda附带了一大批常用数据科学包，方便Python进行数据分析。

#!/usr/bin/python
# -*- coding: UTF-8

'''
DBSCAN算法测试程序
@author: 程序兔
@date: 2018-05-08
'''

from math import *
from sklearn.cluster import DBSCAN


def get_distance(array_1, array_2):
    lon_a = array_1[0]
    lat_a = array_1[1]
    lon_b = array_2[0]
    lat_b = array_2[1]
    radlat1 = radians(lat_a)
    radlat2 = radians(lat_b)
    a = radlat1 - radlat2
    b = radians(lon_a) - radians(lon_b)
    s = 2 * asin(sqrt(pow(sin(a/2),2) + cos(radlat1) * cos(radlat2)*pow(sin(b/2),2)))
    earth_radius = 6378137
    s = s * earth_radius
    return s


def create_coordinate_list():
    result = []
    # 第1个聚类簇，3个点在500米范围内
    point_11 = [116.501146, 39.9915]
    point_12 = [116.501452,39.991002]
    point_13 = [116.501685,39.990491]

    # 第2个聚类簇，2个点在500米范围内
    point_21 = [116.528509,39.995549]
    point_22 = [116.530808,39.99389]

    # 噪音点
    point_31 = [116.419131,40.024949]

    # 第3个聚类簇，2个点在500米范围内
    point_41 = [116.51913,39.972384]
    point_42 = [116.520747,39.972025]

    # 噪音点
    point_51 = [116.53075,39.883273]

    result.append(point_11)
    result.append(point_12)


    result.append(point_21)
    result.append(point_22)

    result.append(point_31)

    result.append(point_41)
    result.append(point_42)

    # 此处加入第1个聚类簇的第3个点
    result.append(point_13)

    result.append(point_51)

    return result


def main():
    # 模拟坐标数据
    coordinate_list = create_coordinate_list()

    # DBSCAN聚类
    dbscan = DBSCAN(eps=500, min_samples=2, metric=get_distance).fit(coordinate_list)

    # 输出[ 0  0  1  1 -1  2  2  0 -1]
    print(dbscan.labels_)


if __name__ == '__main__':
    main()

dbscan.labels_数组就是聚类簇索引列表，噪音点为-1，其他代表各个簇的索引，顺序依次对应坐标点列表。