2.4. Model Building

This section discusses model building with the %Boost macro introduced in Section 2.3.

Each modification of the AdaBoost algorithm (AdaBoost, Real AdaBoost, Gentle AdaBoost and LogitBoost) performs differently on various types of data. Because of this fact, we should evaluate the performance of each model on a training and testing set to identify the model with the best performance. For the permeability data, Program 2.9 performs each version of boosting on the training and testing sets (TRAIN and TEST data sets) created in Program 2.7. The program also computes and plots the associated misclassification errors to gauge the predictive ability of each model.

Example 2-9. Performance of the four boosting algorithms in the permeability ...

Get Pharmaceutical Statistics Using SAS now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.