TFDV Validation only for Categorical?

In Course 2, week1, Validating Data. He mentions that TFDV only supports skew detection for categorical data. What do we do for continuous data?

Hi @Matthew_Avaylon ,

For continuous data you can use the Jensen-Shannon divergence which compares the distributions of numerical data. Here you can set your own thresholds for how similar you want the data distributions to be before triggering alarms:

Hope that helps.
Best regards,
Maarten