Improving the performance of evolutionary algorithms for decision rule learning