Appropriate Order of Feature Selection Techniques Selection based on Filter and Wrapper Approaches

Numerous fields, including the classification of spam emails, cancer cells, fraudulent claims, credit risk, text categorization, and DNA microarray analysis, have extensively used feature selection. However, one of the most significant applications of feature selection technique is in cancer diagnosis through gene expression data. With many genes but few samples, microarray data is a high-dimensional dataset. The purpose of feature selection is to determine how few characteristics are required to provide high classification accuracy. A crucial stage in the analysis of microarray data is feature selection, which lowers the data's dimensionality and boosts the effectiveness of the models. Building predictive models for classification includes predicting the target variable using a variety of input variables (features). This study looks at the best way to apply the filter and wrapper techniques to improve classification results. The computation uses three filters and one wrapper technique, and the results are reported using four performance measures. Three microarray cancer data are used throughout the entire trial. The result analysis shows the order of feature selection in terms of filter and then wrapper techniques outperform.