连接两个数据集，都带有点

Question

我有两个带点的 csv 文件。一个学校数据集（纬度、经度和学校名称）和一个带有房屋坐标（纬度、经度和houseid）的数据集。我想列出距离学校 500 米范围内的所有房屋。我真的不知道如何在 Python 中与 geopandas 进行空间连接。有人可以帮帮我吗？schools.csv56.039484;14.164114;Parkskolan56.029687;14.159337;Centralskolanhouses.csv56.039240;14.165066;156.039008;14.166709;256.038608;14.169420;3

Answer 1

获得解决方案的主要步骤：

1. 将 2 个数据文件读入数据帧

2. 设置 CRS('epsg:4326') 并从 (lat,long) 为两个数据帧创建点几何

3. 对于schools数据帧，将 CRS 转换为 UTMzone 33N

4. 在schools数据帧上做缓冲（半径 = 500m）

5. 在schools数据帧上，执行并将 500m 缓冲区设置为新的geometry

6. 在公共 CRS之间houses和schools中进行适当的空间连接

7. 在houses_joined数据框中获取结果

这是工作代码：

import pandas as pd

import geopandas as gpd

from shapely.geometry import Point, Polygon

# School data

# -----------

# read `schools.csv`, data are in (lat,long); 'epsg:4326'

#

# lat;lon;school_name

# 56.039484;14.164114;Parkskolan

# 56.029687;14.159337;Centralskolan

df_schools = pd.read_csv('schools.csv', na_values=['NaN'], sep=';')

# create Point geometry objects from (lon,lat)

sch_geom = [Point(xy) for xy in zip(df_schools.lon, df_schools.lat)]

# set initial coordinate ref system, and geometry column to the dataframe

gdf_schools = gpd.GeoDataFrame(df_schools, crs={'init': 'epsg:4326'}, geometry=sch_geom)

# convert CRS from (lat,long) to UTMzone 33N

# and get new dataframe: gdf_schools_utm33N

gdf_schools_utm33N = gdf_schools.to_crs(crs="+proj=utm +zone=33 +ellps=WGS84 +datum=WGS84 +units=m +no_defs")

# Note: crs="..." can be replaced by epsg=32633

# do buffering, radius: 500m

gdf_schools_utm33N['buffer_geometry'] = gdf_schools_utm33N.geometry.buffer(500)

# rename `geometry` -> `original_geometry`; `buffer_geometry` -> geometry

# .. and set column `geometry` as the default geometry data of the geodataframe.

gdf_schools_utm33N = gdf_schools_utm33N.rename(

    columns={'geometry':'original_geometry', 'buffer_geometry':'geometry'}).set_geometry('geometry')

# Houses data

# -----------

# read `houses.csv`, data are in (lat,long); 'epsg:4326'

# lat;lon;houseid

# 56.039240;14.165066;1

# 56.039008;14.166709;2

# 56.038608;14.169420;3

# 56.046108;14.171420;4

df_houses = pd.read_csv('houses.csv', na_values=['NaN'], sep=';')

# I add the 4th house that is too far away from all schools

# The 4th house: 56.046108  14.171420   4

# create Point geometry for the houses, and init CRS

hs_geom = [Point(xy) for xy in zip(df_houses.lon, df_houses.lat)]

gdf_houses = gpd.GeoDataFrame(df_houses, crs={'init': 'epsg:4326'}, geometry=hs_geom)

# options: plot the schools' buffers and all the houses

ax = gdf_schools_utm33N.plot(color='lightgray', edgecolor='green', alpha=0.5)

gdf_houses.to_crs(epsg=32633).plot(ax=ax, color='red')

# ******* Spatial Join *****************

# houses data frame needs CRS conversion

hss = gdf_houses.to_crs(epsg=32633)

# do spatial join of houses(points) ~ schools(circles of 500m radius)

houses_joined = gpd.sjoin(hss, gdf_schools_utm33N, op='within', how='inner')

# print out the successful joined rows (house_id, school_names)

# this prints house_id + school_name 

houses_joined[['houseid','school_name']]

# Output: house_id, school_name

# 1    Parkskolan

# 2    Parkskolan

# 3    Parkskolan

结果图：