Enhanced SVM-CRFE-GK integrating Optimized Chi-RFE Feature Selection and Greedy Kernel for COVID-19 Prediction

Abstract

In the Pandemic Situation, Health Care data are accumulating at faster rate, but interpreting, predicting and classifying them is still a challenging one. Machine Learning Models has different algorithms to simplify the problem of interpretation, prediction and classification of data set either it is structured or unstructured. Mainly, for prediction, among many feature selection algorithms simple Support Vector Machine (SVM), Support Vector Machine with Recursive Feature Elimination (SVM-RFE) identifies dependent features and improves the prediction rate and accuracy. But the SVM, SVM-RFE models are not enough when the data size is very huge, unstructured and having high dimensions. To overcome the limitation, this research work propose an Enhanced SVM-CRFE-GK (SVM-Chi-Square-RFE with Greedy Kernel) model which coupled Chi-Square feature selection algorithm to identify the dependent features, RFE to remove the irrelevant features based on the weight vectors with reduced number of iterations without losing the accuracy and uses Greedy Kernel method to optimize performance and to improve the classification accuracy in less time consumption and memory utilization.