Improvement with Chi Square Selection Feature using Supervised Machine Learning Approach on Covid-19 Data

Abstract

The purpose of this study is to make improvements to previous studies with better accuracy than 80.90%. This study will add a Chi-Square selection feature with a supervised machine learning approach, namely the Naïve Bayes algorithm and SVM, to improve accuracy. The test before using the Chi-Square selection feature on the SVM algorithm got 85.56% results, and the Naïve Bayes algorithm got 85.19% accuracy. After adding the selection feature, the SVM algorithm gets an accuracy of 83.86%, and the Naïve Bayes algorithm gets an accuracy of 87.09%. The Chi-Square feature selection method is an algorithm for selecting features, discarding irrelevant features. There are differences in results when testing. SVM, after using the selection feature, the accuracy decreases, while in Naïve Bayes, the accuracy increases. However, when compared to previous studies, our research has improved the accuracy of both classification algorithms, namely SVM and Naïve Bayes.