Track B: quick clarification regarding train/test folders in the runtime repo

PETs Prize Challenge
1

Hi PETs challenge organizers,

For Track B (pandemic forecasting), since the train/test split is split on days and the learning task is to predict individual’s state in the final week (first 56 days for training and last 7 days for testing), does it mean that for a particular scenarioXX/ in the runtime data folder, the train/ and test/ subfolders have exactly the same set of clients (partitions), and each client has (mostly) the same set of CSV files with same content (e.g. persons_df / household_df)?

Please let me know if I missed something, thanks!

2

Hi @kzliu,

That is correct—most of the files are the same between train and test for the pandemic forecasting track. You can take a look at the pandemic-partitioning-example.ipynb on the data download page. This notebook shows some examples of how the files are produced and includes some additional explanation.

1 Like