CY-Bench (Crop Yield Benchmark) is a comprehensive dataset and benchmark to forecast crop yields at subnational level. CY-Bench standardizes selection, processing and spatio-temporal harmonization of public subnational yield statistics with relevant predictors. Contributors include agronomers, climate scientists and machine learning researchers.
@lilybellesweet: This would provide useful context for researchers intending to work with the dataset, and insights gained can be usefully compared with those from previous studies that have assessed this for specific crops and countries. For the camera-ready version, we therefore intend to include an ablation study which compares model performance using varying combinations of weather features, soil data and remote-sensing derived vegetation indices (FPAR, NDVI).
@lilybellesweet: This would provide useful context for researchers intending to work with the dataset, and insights gained can be usefully compared with those from previous studies that have assessed this for specific crops and countries. For the camera-ready version, we therefore intend to include an ablation study which compares model performance using varying combinations of weather features, soil data and remote-sensing derived vegetation indices (FPAR, NDVI).