The Smart Trees module is an extension of the Classification Trees module. The knowledge of the Classification Trees module is required to understand the Smart Trees module. Smart Trees are characterized by quite strong predictive power among classification methods and run effectively on huge data sets with thousands of variables.

Like Classification Trees, Smart Trees have two major functions: predictive and descriptive. Other applications include variable importance analysis.

An important advantage of Smart Trees is that during the model building phase a classification error estimate is automatically obtained as a by-product of the way in which the learning samples are chosen for each tree.

Smart Trees also return more smoothed-out results than single Classification Trees, especially over large data sets, which is due to the fact that the final decision is averaged over the answers provided by many different Classification Trees.

Like Classification Trees, Smart Trees do not have any specific data requirements.

One of disadvantage of the Smart Trees classification method is its relative obscurity. Smart Tree models work like a black-box oracle: there is no clear view of the decision process and model structure is difficult to interpret due to its large size and complexity. Because the output is an average over many independent decisions it is often impossible to explain why the model answered in a particular manner.