| Summary: | Coronavirus Disease (COVID-19) is a global concern as it has spread throughout the world, infecting millions. Those infected were presented with symptoms like fever, cough, fatigue, headache, shortness of breath, sore throat, myalgia, arthralgia, nausea, diarrhoea, chest pain, loss of smell and taste. Although there have been studies carried out regarding this disease, the relationship between the symptoms and the disease severity remains unclear. Few studies have used data mining approaches in classifying COVID-19 severity levels based on symptoms. Therefore, the goal of this study was not only to determine the severity indicators of COVID-19 but also to identify early symptoms of COVID-19 and to model the relationship between COVID-19 symptoms to predict COVID-19 severity levels. The software used in this study for data mining analysis was the Waikato Environment for Knowledge Analysis (WEKA) version 3.8. The data collection which involves two case studies related to COVID-19 were retrieved from Kaggle and a research journal from Turkey. Data pre-processing was carried out to identify and remove outliers. Missing values were treated using filtering and imputation methods. The classification algorithms: J48, SMO, Random Forest, and Simple Logistic were executed and tested to classify data into three classes: mild, moderate, and severe. Results show that symptoms like dyspnoea and breathing problems were the main indicators of severe COVID while those experiencing symptoms like loss of smell were more likely to be categorized under mild or moderate COVID level.
|
|---|