For my use case, i’m wondering how well is DataRobot at understanding the data type of the predictors?
I don't have the details about your use case, but DataRobot automatically detects numerical, categorical, dates, percentages, currencies, lengths, and unstructured text. This is done during the first exploratory analysis referred to as EDA1.
Does this answer your question?
Hi nico1e,
In general, DataRobot automatically detects the type of the feature. Obviously, the specific type of problem / domain will determine if certain numbers, particularly integers, are treated as numerics or categoricals. E.g the number 401 in healthcare could be an ICD9 code, a lab test reading, or a patient ID.
The good news is that you can easily, and in bulk, convert variables to different types manually as needed.
Cheers,
Duncan