Parallel implementation and analysis of the Genetic Rule and Classifier Construction Environment

The Genetic Rule and Classifier Construction Environment (GRaCCE)is an alternative to existing decision rule induction (DRI) algorithms. GRaCCE is a multi-phase algorithm which uses a genetic algorithm based search to reduce the number of features to those that make the most significant contributions to the classification. This feature selection increases the efficiency of the rule induction algorithm. However, feature selection is shown to account for more than 98 percent of the total execution time of GRaCCE on the tested data sets. The primary objective of this research effort is to improve the overall performance of GRaCCE through the application of parallel computing methods to the feature selection algorithm. The implementation of a parallel feature selection algorithm is presented. Experiments employed to test this parallel implementation are discussed followed by an analysis of the results which clearly show that GRaCCE efficiency is improved through the use of parallel programming techniques.