Better decision-making. Faster performance.
Easier than ever.
Try Minitab CART Free

CART introduced several new methods for growing trees, including Gini and Twoing. To cover a broad variety of problems, CART also includes special provisions for handling ordered categorical data and the growing of probability trees.

CART trees deliberately restrict themselves to two-way splits of the data, intentionally avoiding
the multi-way splits common in other methods. These binary decision trees divide the data into
small segments at a slower rate than multi-way splits and thus detect more structure before too
few data are left for analysis. Decision trees that use multi-way splits fragment the data rapidly,
making it difficult to detect patterns that are visible only across broader ranges of data values

CART's developers determined definitively that no stopping rule could be relied on to discover the optimal tree. They introduced the notion of over-growing trees and then pruning back; this idea, fundamental to CART, ensures that important structure is not overlooked by stopping too soon. Other decision-tree techniques use problematic stopping rules that can miss important patterns.

When searching for patterns in data it is essential to avoid the trap of "overfitting", that is, of finding patterns that apply only to the training data. CART's embedded test disciplines ensure that the patterns found will hold up when applied to new data. Further, the testing and selection of the optimal tree are an integral part of the CART algorithm. In other decision-tree techniques, testing is conducted only optionally and after the fact and tree selection is based entirely on training data computations.

For single-variable splitting criteria, CART includes Gini, Twoing, Entropy and Class Probability for classification trees, and Least Squares and Least Absolute Deviation for regression trees. The default Gini method frequently performs best, but by no means is Gini the only method to consider in your analysis. In some circumstances the Twoing method will generate more intuitive trees.