Validating your data

Use the chap validate command to check that your CSV is Chap-compatible before running an evaluation.

Basic validation

chap validate --dataset-csv [YOUR-FOLDER-PATH].csv

e.g chap validate --dataset-csv example_data/laos_subset.csv

This checks for:

• Required columns (time_period, location)
• Missing values (NaN) in covariate columns
• Location completeness (every location has the same set of time periods)

With --model-name, it also checks that the dataset has the covariates and period type the model requires. chap eval warns about these problems and runs anyway; chap eval --strict stops on them.

With --data-configuration, it also checks that no masked range falls in the evaluation window of a backtest, which chap eval refuses. Give the backtest you will run with --backtest-params, and a masked range in its window is an error; in a Chap project, the project's eval settings are used. Without them, chap validate checks the window of chap eval's default backtest and only warns.

Validate GeoJSON (if applicable)

If a GeoJSON file with the same base name sits next to your CSV (for example laos_subset.csv and laos_subset.geojson), it is discovered and loaded by the same command. Locations in the CSV without a matching feature in the GeoJSON are logged as a warning and dropped from the loaded dataset; validation does not fail on them, so check the output for that warning.