Covid-19 Prediction Model Using Data Mining Algorithms

Abstract

From 2019 until today, the whole world is panicking due to the pandemic of the Corona virus or the so-called COVID-19, so it was inevitable to search for a way to predict the disease before it occurs. Disease forecasting requires huge databases, human hands, and high-speed technologies. Due to the rapid spread of this pandemic, scientists have looked at using data mining methods to predict diseases. Predicting and early diagnosis of disease through mining algorithms reduces human errors, saves money and scientists make the most accurate decision. It then shortens the long time needed to detect COVID-19 using available algorithms and tools that rely on data such as lung images or disease symptoms such as temperature, among others. In this research, two data mining algorithms were used to predict COVID-19 that support vector machine (SVM) and naïve bays (NB) using a dataset from the Korea Centers for Disease Control (KCDC). Feature selection was performed using Correlation based Feature Selection (CFS) after previously processed data. Performance measures for the research proved that SVM is best classifier of NB with accuracy, sensitivity and specificity of SVM being 96.72%, 94.08%, and 97.96% respectively. The receiver operating characteristic (ROC) curve also demonstrated better SVM performance than NB for predicting COVID-19.