定制公交优化问题Customized Bus（一）规划中......

场景设置：单出发点多到达点，每个小时的固定时间发车

第一步：模拟一个地区的出行需求，这里设置500个订单需求点（经纬度）并随机分配给各个小时（7到22点）。将这些点聚类，聚类结果设置为定制公交下车点。

下二步预告：根据下车点间距离生成下车顺序，使得总里程最短，设置订单需求的退票率。

优化方向：下车点选择和顺序确定？票价的确定使得总利润最大？设置权重？其实还没想好，如果有大佬看到还望不吝赐教~

# 导入所需包
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import random
import math
from sklearn.cluster import KMeans

# 随机生成500个初始需求及所在位置、需求时间
def generate_random_gps(base_log=None, base_lat=None, radius=None):
    radius_in_degrees = radius / 111300
    u = float(random.uniform(0.0, 1.0))
    v = float(random.uniform(0.0, 1.0))
    w = radius_in_degrees * math.sqrt(u)
    t = 2 * math.pi * v
    x = w * math.cos(t)
    y = w * math.sin(t)
    longitude = y + base_log
    latitude = x + base_lat
    return longitude, latitude
c_lon = list()
c_lat = list()
c_hour=list()
for i in range(500):
    lon, lat = generate_random_gps(base_log=120, base_lat=30, radius=100000)  # 基准点经纬度和距离基准点的半径
    hour=random.randint(7, 22)
    c_lon.append(lon)
    c_lat.append(lat)
    c_hour.append(hour)
    # random.random()
dict = {'lon':c_lon,'lat':c_lat,'hour':c_hour}
df_ini= pd.DataFrame(dict)
df_ini.sort_values(by='hour',inplace=True)

def kmeans():
    df = df_ini[df_ini.hour == 7].reset_index(drop=True)
    feature_location = df[['lon', 'lat']]
    kmean = KMeans(n_clusters=3)  # 三个下车点
    km = kmean.fit(feature_location)
    df['cluster'] = km.labels_
    centers = df.groupby("cluster").mean().reset_index()
    for i in range(8,23): # 每小时聚类一次
        df = df_ini[df_ini.hour == i].reset_index(drop=True)
        feature_location = df[['lon', 'lat']]
        kmean = KMeans(n_clusters=3) # 三个下车点
        km = kmean.fit(feature_location)
        df['cluster'] = km.labels_
        center_s = df.groupby("cluster").mean().reset_index()  # 下车点位置
        centers=pd.concat([centers, center_s],axis=0).reset_index(drop=True)

        # 画图
        '''df1 = df[df.cluster == 0].reset_index(drop=True)
        x = df1['lon']
        y = df1['lat']
        df2 = df[df.cluster == 1].reset_index(drop=True)
        x1 = df2['lon']
        y1 = df2['lat']
        df3 = df[df.cluster == 2].reset_index(drop=True)
        x2 = df3['lon']
        y2 = df3['lat']
        ax1 = plt.axes()
        ax1.scatter(x, y)
        ax1.scatter(x1, y1, color='red')
        ax1.scatter(x2, y2, color='green')
        plt.show()'''

    print(centers)

if __name__ == "__main__":
    kmeans()