MAD vs number of trees plot for Fit Model and Discover Key Predictors with TreeNet® Regression


This command is available with the Predictive Analytics Module. Click here for more information about how to activate the module.

The MAD vs Number of Trees Plot displays the mean absolute deviation on the y-axis and the number of trees on the x-axis. The mean absolute deviation indicates whether the model is a good fit. Use the test results to assess the performance of the model to predict new observations. Compare the training results and the test results to see whether there are overfitting problems with the model for the training data set.

This analysis grows 5000 trees. The optimal number of trees is 4905. The optimal value for the test data when the number of trees is 4905 is approximately 39580.

When the squared error or loss function method determines the number of trees for the optimal model, then Minitab displays the R2 vs Number of Trees plot.


Lower MAD values indicate a better model. The reference line indicates the optimal MAD value for the test data and the number of trees in the model. If the test curve indicates an insufficient model, consider whether to retry the analysis with alternative settings, such as larger or smaller learning rates, or a larger subsample fraction.