In the challenge description, it says the following:
“A panel of judges will review and rank the final submissions in order of priority: accuracy of the solution on a holdout data set using RMSE…”
I want to make sure our model is easy for the judges to evaluate. Is this holdout going to be for different locations than the ones we’ve used for training and predicting so far? Or is it going to be for a different year than the provided test data (i.e. predict 2021 instead of 2020)?