TypeError when writing GeoPandas GeoDataFrame

Question

I have a GeoDataFrame as follows:

import pandas as pd
import geopandas
df = pd.DataFrame([[1, 0], [0, 1]], columns=[['location', 'location'], ['x', 'y']])
df["geometry"] = geopandas.points_from_xy(df['location','x'], df['location', 'y'])
gdf = geopandas.GeoDataFrame(df, geometry="geometry", crs="EPSG:4326")
gdf

#   location                    geometry
#          x  y                         
# 0        1  0  POINT (1.00000 0.00000)
# 1        0  1  POINT (0.00000 1.00000)

Problem: Using the GeoDataFrame to_file method results in a not very explicit TypeError :

gdf.to_file("test.shp")



# ... 
#
# TypeError: Cannot interpret '<geopandas.array.GeometryDtype object at 0x7f99fdf45070>' as # a data type

What could be the reason and how could this be solved?

Note: This error is already described in another question but in an unrelated context.

Answer 1

It appears that GeoDataFrame.to_file requires a column with exactly the name geometry : No multilevel columns, no geom , no geometry_ , etc.

Therefore, the above problem can be solved by flattening the column index:

gdf_flatcols = gdf.set_axis([x[0] if x[1] == "" else "_".join(x)
                             for x in gdf.columns], axis=1)

and to_file will work just fine:

gdf_flatcols.to_file("test.shp")