Notebooks/analysis

[tommylees112]

Producing the final notebooks as part of the ECMWF Summer of Weather Code.

Split into just two notebooks: 1) Pipeline.ipynb

• an overview of the pipeline and its functionality
• the diagrams used from before
• a relatively thorough overview of each component part of the pipeline

2) Analysis.ipynb

• the play along notebook that helps us to demonstrate the analysis that can be done with the pipeline results.

Requires:

1. A zip file (data/zip_data.zip)
2. How to share this data? Can we use GoogleDrive? Too big too share via github unfortunately (~1GB)
3. We need the trained model.pt file (my one doesn't seem to produce our sota RMSE (i get an rmse ~40)
4. need help with the shap values. need to be computed so that can be plotted (using the static cell from @gabrieltseng's draft notebook

[gabrieltseng]

Nice work Tommy.

Some of the notebooks seem incomplete / don't run? (Specifically, 18_tl_analysis_chirps_grid.ipynb). It's probably fine, since its in the drafts, but is this a bug in code which should be fixed?

[tommylees112]

okay i think i fixed that bug!

[gabrieltseng]

Nice ! Could you rerun that notebook too (just prevents buggy notebooks from being uploaded)

[gabrieltseng]

LGTM once that happens