Are boxes better for classification?

The design of a classifier usually has the important step of attribute or feature selection. A computationally tractable scheme almost always relies on a subset of attributes that optimize a certain criterion is chosen, resulting in a good suboptimal solution. We show that it is possible to directly define the region for each class in terms of an n-dimensional box, using all numeric attributes (discrete or continuous). Several shapes are used for the modeling of classes. Shapes are tested and results compared. As with our previous algorithm, this classifier is transparent. The approach compares favorably with previous approaches in both accuracy and efficiency. It has the added advantage of being able to classify when one class is entirely within another class