Project trashpandas #51

New Issue

phryk · 2021-03-05T08:21:36Z

phryk commented

2021-03-05 08:21:36 +00:00

This has been a long time coming…

Rip out pandas – it's way too horrible in regards to [de]serialization.

Keep current [de]serialization, but simplify where appropriate.

Implement own analysis system on top of numpy, shapely and possibly pyproj.
Implement functionality that we used from pandas by now:

JOIN
>
>=
==
<=
<

This has been a long time coming… Rip out pandas – it's way too horrible in regards to [de]serialization. Keep current [de]serialization, but simplify where appropriate. Implement own analysis system on top of `numpy`, `shapely` and possibly `pyproj`. Implement functionality that we used from pandas by now: * JOIN * \> * \>= * == * <= * <

phryk added this to the 0.0.0.0-alpha milestone 2021-03-05 08:21:36 +00:00

phryk added the

labels 2021-03-05 08:21:36 +00:00

phryk commented

2021-05-13 11:04:45 +00:00

It is done.

We now have fuzzy matching on joins including levenshtein distance for strings and much more complex filtering than before including all shapely binary predicates for geodata with the possibility to filter against manually built polygons.

Some places are still lacking validation, new bugs have probably been introduced.
Still feels more comfortable than trying to throw hoops around pandas. ¯\_(ツ)_/¯

It is done. We now have fuzzy matching on joins including levenshtein distance for strings and much more complex filtering than before including all [shapely binary predicates][1] for geodata with the possibility to filter against manually built polygons. Some places are still lacking validation, new bugs have probably been introduced. Still feels more comfortable than trying to throw hoops around pandas. ¯\\\_(ツ)\_/¯ [1]: https://shapely.readthedocs.io/en/stable/manual.html#binary-predicates

phryk closed this issue

2021-05-13 11:04:45 +00:00

phryk referenced this issue

2021-05-13 11:11:15 +00:00

GeoData analysis #6

phryk referenced this issue

2021-05-13 11:15:10 +00:00

Analysis: First steps #3

Sign in to join this conversation.