How eco4cast/forecast_submissions works
EFI Theory is a collection of ecological forecasting models for multiple NEON data streams in the EFI forecast challenge. Its documented operational workflow is that forecast models run daily; the README also describes a separate train-then-forecast approach for computationally intensive machine-learning models. The tree shows many model-family directories and a submissions-analysis area, but no source excerpts were provided, so internal calls, data transfers, and submission handoffs cannot be verified. Model groups below represent the repository’s documented capabilities and directory structure, not a confirmed execution pipeline.
Components 17
Time-series models
- ARIMA
Generate_forecasts/ARIMA - ETS
Generate_forecasts/ETS - TBATS
Generate_forecasts/TBATS - NNETAR
Generate_forecasts/NNETAR - Temperature linear models
Generate_forecasts/temp_lm - All-site linear models
Generate_forecasts/temp_lm_all_sites
Theory-group models
- Theory-group forecasts
Generate_forecasts
Machine-learning models
- SVM
Generate_forecasts/tg_SVM - XGBoost
Generate_forecasts/tg_XGBoost - Lasso
Generate_forecasts/tg_lasso - Random forest
Generate_forecasts/tg_randfor - Bagged MLP
Generate_forecasts/tg_bag_mlp - Bagged trees
Generate_forecasts/tg_bag_tree_all_sites
Forecast analysis
- Submission analysis
Analyze_submissions
Other components
- EFI forecast challenge
- NEON data streams
- Daily forecast models
Connections 2
- Daily forecast models → EFI forecast challenge: run forecasts
README.md - NEON data streams → Daily forecast models: forecast across