Abstract
In the paper, we study an extension of dynamic programming approach which allows optimization of approximate decision rules relative to the number of misclassifications. We introduce an uncertainty measure J(T) which is a difference between the number of rows in a decision table T and the number of rows with the most common decision for T. For a nonnegative real number gamma, we consider gamma-decision rules that localize rows in subtables of T with uncertainty at most gamma.
The presented algorithm constructs a directed acyclic graph Delta(gamma)(T). Based on this graph we can describe the whole set of so-called irredundant gamma-decision rules. We can optimize rules from this set according to the number of misclassifications. Results of experiments with decision tables from the UCI Machine Learning Repository are presented.