■suumoの物件分析 #1 「関東の家賃偏差値を求めて地図上にヒートマップとして表示する」

conda install -c conda-forge geopandas

conda install descartes

import numpy as np
import pandas as pd
import geopandas as gpd
import matplotlib.colors
import matplotlib.pyplot as plt
import re
import statistics
import math

from matplotlib.colors import Normalize
from matplotlib.cm import ScalarMappable, get_cmap

### (1) データ側の処理
filename = "./kanto_map/kantou.geojson"
df_yachin = pd.read_excel('./kanto_manshon_oneroom.xlsx')

def return_yachine_value(v):
    if v != "-":
        return float(re.match("\d.\d", v).group())*10000
    elif v == "-":
        return "-"

def return_place(row):
    #print("test")
    if ( row['N03_003'] is not None ) and (re.match(".*郡",row['N03_003'])): # ●●郡の場合
        #print(row['N03_003'])
        return row['N03_003']
    elif ( row['N03_003'] is not None ) and (re.match(".*市",row['N03_003'])): # ●●市の場合
        return row['N03_003'] + row['N03_004']
    else:
        return row['N03_004']
    
def remove_specified_values(arr, value):
    while value in arr:
        arr.remove(value)
        
def get_hensa(yachine, average, pstdev):
    if yachine != "-":
        return 50 + 10*((yachine - average)/pstdev)
    elif yachine == "-":
        return "-"
    
df_yachin["yachin_value"] = df_yachin["家賃"].apply(return_yachine_value)
yachin_list = []
yachin_list = df_yachin["yachin_value"].values.tolist()

remove_specified_values(yachin_list, "-")

yachine_average = statistics.mean(yachin_list)
yachine_pstdev = statistics.pstdev(yachin_list)
df_yachin["家賃偏差値"] = df_yachin["yachin_value"].apply(get_hensa, average=yachine_average, pstdev=yachine_pstdev)

### (2) マップ側の処理
def colors_scale(arr):
    n_min = arr.min(skipna = True)
    n_max = arr.max(skipna = True)
    cmap = plt.cm.jet
    
    norm = Normalize(vmin=n_min, vmax=n_max)
    arr = [(1.0, 1.0, 1.0, 1.0) if np.isnan(r) else cmap(norm(r)) for r in arr]
    
    return arr, cmap, norm

df_map = gpd.read_file(filename, encoding='SHIFT-JIS')

map_hensa_list = []
df_map["place"] = df_map.apply(return_place,axis=1)

for index_1, row_1 in df_map.iterrows():
    match_flag = 0
    for index_2, row_2 in df_yachin.iterrows():
        if row_2['市区町村'] == row_1['place']:
            if row_2['家賃偏差値'] != "-":
                #print(row_1['市区町村'])
                map_hensa_list.append(row_2['家賃偏差値'])
                match_flag = 1
                #print(map_hensa_list)
            elif row_2['家賃偏差値'] == "-":
                map_hensa_list.append(np.nan)
                match_flag = 1
    if match_flag == 0: ## N03_004の市町村がsuumo側の市町村に無かった場合
        map_hensa_list.append(np.nan)
        
df_map["家賃偏差値"] = map_hensa_list

target_color, cmap, norm = colors_scale(df_map["家賃偏差値"])

fig, ax = plt.subplots(figsize=(20,12))
df_map.plot(ax=ax, color=target_color, edgecolor='black')
plt.xlim([138,141])
plt.ylim([34.8,37.3])

sm = ScalarMappable(cmap=cmap, norm=norm)
sm._A = []
cbar = plt.colorbar(sm)

ticks = np.linspace(norm.vmin, norm.vmax, 10)
cbar.set_ticks(ticks)
cbar.ax.set_yticklabels([str(int(s)) for s in ticks])

plt.xticks([])
plt.yticks([])
plt.show()

def return_place(row):

arr = [(1.0, 1.0, 1.0, 1.0) if np.isnan(r) else cmap(norm(r)) for r in arr]

plt.xlim([138,141])
plt.ylim([34.8,37.3])