Why is my output is in Series after creating LineStrings from grouped ID?

Question

I want to group the points by userid then create a LineString for each userid. Here's the GeoDataframe:

My goal is to store the results in a Geodataframe. Here's the code which I actually obtained from one question in this forum several years ago:

line_strings = df.groupby("userid")["geometry"].apply(lambda x: LineString(x.tolist()) if x.size > 1 else x.tolist())

However, the output is in Series instead.

So if I want to convert it back to GeoDataFrame, I unable to do so because it said TypeError: Input must be valid geometry objects: [<POINT (900800.126 7174374.539)>]

So I want to know where is my mistake?

P.S.: I also want to know how can I exclude User IDs that has only one point? (Hence LineString can't be created for these User IDs)

Answer 1

The typical way to "group by" a GeoDataFrame is to use dissolve. You group by one or more columns, and the geometry column of the resulting GeoDataFrame will contain a union of all grouped values.

Next, you can convert the unioned points (or rather multipoints) to linestrings.

Code sample:

import geopandas as gpd
import shapely
from shapely import Point

input = gpd.GeoDataFrame(
    {
        "userid": [1, 2, 2, 3, 3], 
        "geometry": [
            Point(0,0), Point(1,0), Point(1,1), Point(2,0), Point(2,1)
        ],
    },
    crs="EPSG:31370",
)

diss = input.dissolve(by="userid")
diss_multi = diss.loc[diss.geometry.type == "MultiPoint"]

print(diss)

lines = diss_multi.set_geometry(
    diss_multi.geometry.apply(lambda x: shapely.LineString(shapely.get_coordinates(x)))
)

print(lines)

Result:

userid
2       LINESTRING (1 0, 1 1)
3       LINESTRING (2 0, 2 1)

PS: it is always easier if you include a runnable minimal reproducible example in code including input data. Based on how I understood your question I tried to create "runnable" sample data in the code sample.