Mastering Machine Learning with R - Second Edition

Model selection

Recall that our primary objective in this chapter was to use the tree-based methods to improve the predictive ability of the work done in the prior chapters. What did we learn? First, on the prostate data with a quantitative response, we were not able to improve on the linear models that we produced in Chapter 4, Advanced Feature Selection in Linear Models. Second, the random forest outperformed logistic regression on the Wisconsin Breast Cancer data of Chapter 3, Logistic Regression and Discriminant Analysis. Finally, and I must say disappointingly, we were not able to improve on the SVM model on the Pima Indian diabetes data with boosted trees.

As a result, we can feel comfortable that we have good models for the prostate ...