Complexity of Classiication Problems and Comparative Advantages of Combined Classiiers

We studied several measures of the complexity of classiica-tion problems and related them to the comparative advantages of two methods for creating multiple classiier systems. Using decision trees as prototypical classiiers and bootstrapping and subspace projection as classiier generation methods, we studied a collection of 437 two-class problems from public databases. We observed strong correlations between classiier accuracies, a measure of class boundary length, and a measure of class manifold thickness. Also, the bootstrapping method appears to be better when subsamples yield more variable boundary measures and the subspace method excels when many features contribute evenly to the discrimination.