使用 python 根据邻居填充几何线串数据

Question

我正在研究流量检测深度学习模型，我仍在数据预处理中，我的数据位于 .shp 和 .dbf 和 .prj 文件中，现在所有必要的数据都位于 .geojson 文件中我对其进行投影我看到了地图（我们到阿拉斯加）我正在研究，但几何和

shape_leng

列中仍然存在一些缺失值，因为它们相关，这里是非缺失值行的示例：

"Shape_Leng": 7.2826369182000004e-05 }, "geometry": { "type": "LineString", "coordinates": [ [ -85.393274186354574, 31.208875440549207 ], [ -85.393254553255019, 31.208945570581648 ] ] }

这是缺失值行：

"Shape_Leng": 0.0 }, "geometry": null

我想用接近其邻居的值来填充那些 null 和 0.0
请注意，文件中有 6840 个连续的 null 和 0.0 值（其中只有一行，然后在 not null 中），总共有 18354 个缺失值

我尝试了这段代码：

import geopandas as gpd
import pandas as pd
import shapely
from shapely.geometry import LineString

from geopy.distance import geodesic


df = gpd.read_file("geoformat.geojson")


df.reset_index(drop=True, inplace=True)


def interpolate_geometry(row):

     prev_index = row.name - 1
     next_index = row.name + 1


     if pd.notnull(row['geometry']):
         return row['geometry']


     if isinstance(row['geometry'], shapely.geometry.MultiLineString):
         coords = []
         for line in row['geometry']:
             coords.extend(line.coords)
         return LineString(coords)


     if isinstance(row['geometry'], shapely.geometry.LineString):
         return row['geometry']


     prev_coords = df.at[prev_index, 'geometry']
     if not isinstance(prev_coords, shapely.geometry.MultiLineString):
         prev_coords = df.at[0, 'geometry'] if prev_index < 0 else df.at[prev_index, 'geometry']


     if next_index == len(df):
    
         return prev_coords

     next_row = df.iloc[next_index]
     next_coords = next_row['geometry'].coords[0] if next_row['geometry'] else None


     if next_coords is None:
    
         interpolated_coords = prev_coords
     else:
         if isinstance(prev_coords, shapely.geometry.LineString):
             if isinstance(next_coords, shapely.geometry.Point):
                 next_coords = LineString([next_coords])
             elif not isinstance(next_coords, shapely.geometry.LineString):
                 raise TypeError('next_coords must be a LineString or Point object')

         interpolated_coords = LineString([prev_coords, next_coords])

     return interpolated_coords



interpolated_geometries = []


for idx, row in df.iterrows():
    interpolated_geometries.append(interpolate_geometry(row))


df['geometry'] = interpolated_geometries


def interpolate_shape_leng(row):
    if row['Shape_Leng'] != 0.0:
        return row['Shape_Leng']

    
    prev_index = row.name - 1
    while pd.isnull(df.at[prev_index, 'Shape_Leng']) or df.at[prev_index, 'Shape_Leng'] == 0.0:
        prev_index -= 1


    next_index = row.name + 1
    while pd.isnull(df.at[next_index, 'Shape_Leng']) or df.at[next_index, 'Shape_Leng'] == 0.0:
        next_index += 1


    prev_coords = df.at[prev_index, 'geometry'].coords[-1]
    next_coords = df.at[next_index, 'geometry'].coords[0]
    distance = geodesic(prev_coords, next_coords).meters


    prev_shape_leng = df.at[prev_index, 'Shape_Leng']
    next_shape_leng = df.at[next_index, 'Shape_Leng']
    interpolated_shape_leng = prev_shape_leng + (next_shape_leng - prev_shape_leng) * (row.name - prev_index) / (
            next_index - prev_index)

    return interpolated_shape_leng


df['Shape_Leng'] = df.apply(interpolate_shape_leng, axis=1)


df.to_file("geoformat_imputed.geojson", driver='GeoJSON')

这是我得到的错误：

Traceback (most recent call last):
  File "C:\Users\youssef\Desktop\prj.py", line 68, in <module>
    interpolated_geometries.append(interpolate_geometry(row))
                                   ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\youssef\Desktop\prj.py", line 56, in interpolate_geometry
    raise TypeError('next_coords must be a LineString or Point object')
TypeError: next_coords must be a LineString or Point object