Classification Algorithms Based on Linear Combinations of Features

We provide theoretical and algorithmic tools for nding new features which enable better classiication of new cases. Such features are proposed to be searched for as linear combinations of continuously valued conditions. Regardless of the choice of classiication algorithm itself, such an approach provides the compression of information concerning dependencies between conditional and decision features. Presented results show that properly derived combinations of attributes, treated as new elements of the conditions' set, may signiicantly improve the performance of well known classiication algorithms, such as k-NN and rough set based approaches.