MLinPL 2023 - feedback and ideas!

[Calychas]

Hello everyone!
We would like to warmly invite you to share feedback from the tutorial and some ideas for SRAI usage:

• Do you have any feedback regarding the tutorial?
• What are your use cases?
• Could SRAI help you solve your problems?
• Do you have any tips?

We would love to hear about them to include them in future SRAI development.

Thanks, everyone, for the contribution. It is greatly appreciated!

[JakubCha]

Hello guys!

Do you have any feedback regarding the tutorial?

It was intensive. I learned a lot regarding the possibilities that srai gives to present geospatial engineers so it was time well spent!

What are your use cases?

I tend to use more imagery (satellite, aerial) data so it would be nice to see them included in your library, connected with present features for vector data.

Could SRAI help you solve your problems?

Definitely. I will try to use srai in an ML pipeline.

Do you have any tips?

If I had to change anything regarding tutorial I would focus more on srai functionalities, less on (geo)pandas hustle and bustle. At the same time, I know that your focus was also to "introduce geospatial in ML" as a domain so it's hard to balance and make everything in a limited timeframe.

Thank you!

[pmallas]

Howdy SRAI folks!

Thanks for a great library.

Do you have any feedback regarding the tutorial?
Well, it was great. Very in depth and covered a lot of material.

In the answers of the bike example, I believe there was a typo:

# Mark remaining regions as negative
negative_samples = regions_gdf.copy()[~regions_gdf.index.isin(positive_samples_new.index)]

There is an "_new" appended to the DataFrame name.

That of course is minor. But the next cell fails for me and appears to be a bug:

embeddings = embedder.transform(
    regions_gdf=regions_gdf, features_gdf=features_gdf, joint_gdf=joined_features
 )

....

File [~\Miniconda3\envs\ais\lib\site-packages\srai\neighbourhoods\_base.py:154](http://localhost:8891/lab/tree/~/Miniconda3/envs/ais/lib/site-packages/srai/neighbourhoods/_base.py#line=153), in Neighbourhood._handle_center(self, index, distance, neighbours, at_distance, include_center_override)
    152 else:
    153     if at_distance:
--> 154         neighbours.discard(index)
    155     else:
    156         if include_center:

AttributeError: 'list' object has no attribute 'discard'

Thanks
Paul

[gegen07]

See the issue #431!

[pmallas]

Ahh, thanks. I should have looked harder.

[Calychas]

Heyo! This will be resolved in 0.7.1 - #437