Simple empirical forecasts for the VERA forecasting challenge (https://www.ltreb-reservoirs.org/)
- asl.auto.arima
- asl.ets
- asl.tbats
- asl.temp.lm
- asl.met.lm
- asl.met.lm.step
- Figure out if we want to add auto_adam, nnetar
- Load inflow targets
- Copy an existing model folder (I suggest
ARIMA
for a time series model ortemp_lm
for a model with meteorology) - Update
forecast_model.R
to include your new model (most of the script should be able to stay the same) - Update the the file paths in
rerun_forecasts.R
andrun_forecast.R
to source the correctforecast_model.R
script - Run everything in your
forecast_model.R
file, then open.R/generate_tg_forecast
and run the lines of this function individually to make sure everything works. - If that worked, you should be pretty much set up! To create automation, go to
.github/workflows
and copy one of thedo_prediction_XXX.yml
files. Update the model name and file path for your new model. - Push changes and create automation