Roadmap

[emmamendelsohn]

This is roughly the order I imagine addressing things. I wouldn't spend too much time on the data before the model. You can bring in additional variables after the fact. This road map includes all issues in this repo except for those labeled as enhancements. (I suggest reviewing the enhancement issues for ideas about future steps as well as current limitations.)

Please make sure you read the linked issues all the way through, as some are old threads where we have changed course along the way.

Priority Tasks

Preview Give feedback

1. Run data download and processing steps #92
   data-engineering +1
2. Create a first version of the model #95
   model-development +1
3. Model reporting for iterative comparisons #5
   0 of 4
   reporting +1
4. Finalize static data ingest #20
   data-engineering +1
   
5. WAHIS outcome variable - filter out wildlife #90
   bug data-engineering +2
6. Additional data layers and feature engineering #94
   data-engineering +1
7. Revisit data split approach for training and validation #82
   model-development +1

Prediction and automation tasks

Preview Give feedback

1. Create prediction pipeline #93
   predictions +1
2. Set up containerization and GH action to include grib system tools #7
   infrastructure +1
   
3. Convert containerTemplateUtils to s3fs #89
   infrastructure maintenance +2

Issues that will arise when moving beyond RSA

Preview Give feedback

1. MODIS NDVI rasters missing eastern Africa #72
   bug maintenance +2
2. Clip augmented dataset to continent boundaries #75
   data-engineering maintenance +2