Feature Selection using Genetic Algorithm for Microarray Data Classification

Abstract

Microarray datasets are critical in detecting cancer, tumor, and various other diseases. The problem with the microarray dataset is that it has more features compared to the samples and it affects the performance of the algorithm. So, to get exact information from the microarray dataset, we need a hefty or robust technique. So here comes the feature selection technique in the picture. We will use a Genetic Algorithm for feature selection in combination with the classification algorithms Random Forest, Decision Tree, and Logistic Regression. So, a minimal number of features will be selected from a large number of features using this and will be used to find the accuracy. We have tested this on five microarray datasets Lung, Breast, Lymphoma, Ovarian, and CNS. We have seen that feature selection using genetic algorithm has show that our accuracy has been increased in all the classifiers.