maps by keyword

[kdutia]

Streamlit app to let users generate maps based on keyword/regex searches.

I had to commit some map projections as there's no easy way of accessing them via code otherwise (I thought about S3 but realised they'd probably be more sensible to keep here as self-contained examples will be beneficial when the repo is public).

Also some commit history weirdness going on with the document viewer being in this PR (also merged to main, which might've been the result of me pushing a local commit..?) but feel free to ignore that.

Screen.Recording.2024-10-09.at.09.44.17.mov

[linear]

PODS-1824 build maps notebook that loads from huggingface

relative and absolute paragraph count per country of:

• all keywords in OR arrangement
• each of the keywords

also return total num geographies and documents

based on explore_keywords function

extensions:

• also count per document; per family
• look at bigram count based on a keyword

notes from Anne

• use this file for geopandas - code is here
• geopandas used in dsci research will throw an error

todo list

• get basic plot working
• return total number of geographies
• return total number of families
• show relative and absolute frequency plots
• show plots per keyword

[AnneIsARealProgrammerNow]

A few commented out lines (including a TODO) which might be deleted but looks good otherwise!