An Ensemble Model using Genetic Algorithm for Feature Selection and rule mining using Apriori and FP-growth from Cancer Microarray data

Several dimension reduction techniques for microarray data have been developed over the years to get differentially expressed genes. However, a very little consensus in their resultant feature subsets for a particular dataset has been noticed. Therefore, to address the aforesaid issue an ensemble of feature selection technique is proposed in this paper. The ensemble is a combination of well balanced collection of filter and wrapper based feature selection methods. For further refinement of the resultant output of ensemble, we have taken genetic algorithm in pipeline to produce a non-local set of robust feature subset. An extensive computational experiment has been carried out on prostate cancer dataset for validation of the method. In addition, we have compared the performance of our method with group genetic algorithm (GGA). Finally, the resultant feature subsets of GA, GGA, and other constituents of ensemble in standalone mode are used for uncovering frequent patterns based on two popular association rule mining like Apriori and FP-growth. The experimental results confirm that the proposed method gives a stable result and is very effective in comparison to GGA for attribute clustering in the selection of relevant features.