Shapely VS geopandas

https://github.com/shapely/shapely First paragraph of readme agrees w/ me

The frames' vector data is generated using a sketch made with my vsketch framework (it involves loading GeoJSON data, processing it with Shapely, and projecting it in 3D with NumPy).

As has been said the definition of a GIS dev is far from written in stone, but to chime in from a personal standpoint: most of what I do is data wrangling/analysis with shapely/geopandas for vectors (or pygeos / fiona for performance when data volumes get large, but seeing as Shapely 2.0 just got released one can likely skip this part) + rasterio for rasters as well as parallelising these tasks for performance if needed (ray is great for that) and then performing machine learning learning against the data (mostly with sklearn and torch).

geopandas and the underlying geometry library shapely provide an alternative to the majority of the basic GIS functionality of qgis. See e.g. https://autogis-site.readthedocs.io/en/latest/ for a tutorial for the basics of these libraries.

Interesting! Full honesty I don’t know, lol. But if I were to guess… something to do with implementation of pygeos and indexing. Recent geopandas versions I believe handle joins with shapely vs. pygeos differently. Maybe if there is a declared predicate, it treats the spatial index differently. https://github.com/geopandas/geopandas/pull/1421

From the little I've seen, Polars looks good, but I'm sticking with Pandas for now... I do a lot of work with GeoPandas. When they release a usable version of GeoPolars, I'll take a look.

Details on the root of the issue.[0]

[0] https://github.com/geopandas/geopandas/issues/2382

This means that almost all dependencies of GeoPandas now support Python 3.11, only Fiona is left (but being worked on). See this tracking issue: https://github.com/geopandas/geopandas/issues/2592

Best python package alternative in my opinion: geopandas

Geopandas uses pygeos to implement fast spatial joins (source code is here for the sjoin function: https://github.com/geopandas/geopandas/blob/main/geopandas/tools/sjoin.py). More from GeoPandas documentation: “GeoPandas offers built-in support for spatial indexing using an R-Tree algorithm. Depending on the ability to import pygeos, GeoPandas will either use pygeos.STRtree or rtree.index.Index. The main interface for both is the same and follows the pygeos model.”