How to use Regression Learner App with NaN values?

I have a table with 3 inputs and an output and I'm using Regression Learner app to train a regression model and so far everything is ok, then I should decrease the measurements of a single input predictive variable and see how the performance changes, so I delete some measurements in a column and therefore they will assume NaN values, but subsequently Regression Learner app don't make predictions for the lines that contain NaN values, how can I solve?

Answer 1

[Fabrice Lambert]

Hi,

As I understand it, the linear regression, SVM, and some GPR algorithms will ignore rows that contain NaNs. Use the tree algorithms, these will ignore only the NaN values in its column.