How to groupby by geometry column with Python?

Question

I'm wondering whether someone can help me with this, may be naive, issue, please? Thanks in advance for your opinion. Q: How can I use groupby to group by ['id', 'geometry']? Assuming the geopandas data reads for: pts =

       id       prix  agent_code          geometry
  0    922769  3000 15  POINT (3681922.790 1859138.091)
  1   1539368  3200 26  POINT (3572492.838 1806124.643)
  2    922769  50   15  POINT (3681922.790 1859138.091)
  3   1539368  200  26  POINT (3572492.838 1806124.643)

I have used something like this:

 pts = pts.groupby(['id', 'geometry']).agg(prom_revenue=('prix',np.mean))..reset_index()

However Python raises the following error:

 TypeError: '<' not supported between instances of 'Point' and 'Point'

Thanks for your help, dudes!

You could implement comparison for the Point class: https://stackoverflow.com/q/5824382/8878627 — S P Sharan, May 03 '22 at 16:29

score 3 · Answer 1 · answered May 03 '22 at 16:40

3

Use to_wkt from geometry column to convert shape as plain text:

out = pts.groupby(['id', pts['geometry'].to_wkt()], as_index=False) \
         .agg(prom_revenue=('prix', np.mean))
print(out)

# Output
        id  prom_revenue
0   922769        1525.0
1  1539368        1700.0

answered May 03 '22 at 16:40

Corralien

109,409
8
28
52

Thanks Corralien. I'm new into geostuffs and I didn't know about the wkt, neither the as_index=False! – Gohann May 04 '22 at 13:01

BeRT2me · Answer 2 · 2022-05-03T17:29:42.637

3

If you take Corralien's answer a step further, you can restore the points again like this:

out = pts.groupby(['id', pts['geometry'].to_wkt()]).agg(prom_revenue=('prix', np.mean)).reset_index()
out.columns = ['id', 'geometry', 'pro_revenue']
out['geometry'] = gp.GeoSeries.from_wkt(out['geometry'])
out = gp.GeoDataFrame(out)
print(out)

Output:

        id                         geometry  pro_revenue
0   922769  POINT (3681922.790 1859138.091)       1525.0
1  1539368  POINT (3572492.838 1806124.643)       1700.0

edited May 03 '22 at 17:29

answered May 03 '22 at 17:12

BeRT2me

12,699
2
13
31

BeRT2me, it was amazing your complimentary answer since I didn't know how to recover back my geometry Points again! That's is what I ned to use then. – Gohann May 04 '22 at 13:02

score 0 · Answer 3 · answered May 19 '23 at 15:46

In the most recent version of geopandas (0.13.0) the problem is now solved, so if you are reading this, you just have to upgrade the package to the latest version and do:

pts = pts.groupby(['id', 'geometry']).agg(prom_revenue=('prix',np.mean)).reset_index()

How to groupby by geometry column with Python?

3 Answers3