Can I identify the extent of overfitting using JMP Pro models built using training, testing and validation data?
May 26, 2016 10:23 AM
| Last Modified: Aug 18, 2020 1:51 PM
In her Mastering JMP webcast, Analyzing and Predicting System Failures, Ghazal Falahi explains that the first thing to do to determine overfitting is to compare the R-squared values of test and training sets. If they are close to one another, the model fit is probably good.. High R-squared for training with low R-squared for test probably indicates overfitting.