I have an LGBM model that trains on machine learning features and predicts some numerical outcome with new data coming periodically. I would like to make sure that the new input data is "similar" with the data that I have trained my model on, in terms of mean. Is there a way of creating a range of expected values (for each feature) so that I am confident that if the new value of the feature falls withing that range, it means the mean of the feature didn't change (significantly)?
Asked
Active
Viewed 16 times
