As we have seen, there are a number of differences between algorithms. For instance, some use the formula notation and some the matrix notation. The caret
package uses a similar notation for all the algorithms it supports. Further, it contains tools that perform sampling operations, such as generating training and testing data with the same characteristics (stratified sampling), the use of boosting of bagging with several algorithms, and cross-validation samples. Examples of cross-validation include, for instance, the use of 10 subsamples, of which one is iteratively used as testing data and the rest as training data (or the leave-one-out cross-validation, where one observation is iteratively used as testing data and the rest as training data). Other features are included as well, such as examining the performance of the classification, as we have done previously. The caret
package will be further discussed in Chapter 14, Cross-validation and Bootstrapping Using Caret and Exporting Predictive Models Using PMML.
3.141.28.107