Feature Selection of Multi-class Data Sets Based on Enhanced Binary Gray Wolf Algorithm and Ant Colony Optimization Algorithm

Feature selection in data exploration and analysis problems for multiclass data sets seeks to select a subset of variables or features that define the data to acquire the most appropriate and concise representation of the available information. This issue can be solved using feature selection, which removes extra and irrelevant data. Computation time, training accuracy, and understanding of training models and resources are all decreased as a result. The previous method Enhanced Cuckoo Search with Ant Colony Optimization algorithm (ECS–ACO), shows slow search speed and low collection accuracy of data features. The improved Binary Grey Wolf algorithm (EBGWO–ACO), ant colony-based optimization, which is the basis of the proposed feature selection method for multi-class datasets, is used to determine the minimal feature set selection. Data is gathered from the standard repository in the first step and start with data pre-processing is based on a filtering process to reduce the missing values or irrelevant values from the dataset. Then, cooperative features are selected from the clustering method for the grouping feature dataset. To identify the cluster of different data points into group or dataset values of help to similar data features. Third step is feature selection is based on the Enhanced Binary Grey Wolf algorithm with Ant Colony Optimization (EBGWO–ACO) is for using analysis the minimum subset of samples and the grey wolf algorithm is used to identify the best features with ACO is based on the similar feature location. By using k5-fold validation against the optimal classifier model, the KNN classification accuracies of the investigated approaches were confirmed, and in terms of classification success, the classification. The findings validate that when the number of characteristics is decreased, the approach is appropriate and classically performs improved.