null condition for test data

[bachlaw]

Thank you for the wonderful package. However, softbart and softbart_regression both seem to become upset if I do not supply test data in addition to the training data. Although I often will supply testing data with training data, sometimes I would rather predict it later or be content to fit the training data only. I definitely do not want to have fitting take longer to fit the same data twice, if that is what is occurring.

Is there a particular command or null/zero matrix that can be passed to allow the functions to run with training data alone? Or is it better to randomly sample a few rows from the training data into a faux testing set and then just ignore it?

Thanks again.

[theodds]

You could either (i) pass in the training data against as test data or (ii) just extract a couple of rows if you are worried about getting slowed down. The cost of evaluating on the test data is small relative to training the model.