COVID-19 risk factors specification using Decision Tree based on the degree of redundancy between features

Abstract

Based on the latest diseases which spread in the whole world and need to be predicted and classified. In addition, when testing and examining the samples will be safer with far data collecting such as COVID-19 cases. Therefore; this research provides a safe and accurate data mining prediction system to make a decision with high performance to prevent this spread. Such a study prevents or at least reduces the effect of contacting suspicious patients with others by providing a discovery system to detect this disease in these samples. Also, this study will reduce the effects of COVID-19 on marketing, teaching, and other different business, which lead to holding this disease separated at home with high knowledge of some symptoms that will be studied to specify the most affected features on this classification. However, this study could provide some information about viruses moving and keeping away at home with an early prediction. In this study, three techniques are applied for 1486 patients after data preprocessing and preparing for the performance evaluation. Risk factors are determined using a features selector and study of the effect of these features before and after minimization on the whole proposed model. Differences and reasons are shown in this paper due to different results which occurred while omitting unnecessary data. All the proposed models showed an enhancement in their performances after selecting the most affected features. But, DT showed the best prediction accuracy with about 96% compared to other models. On the other hand, other parameters are explained and showen some more advanced in the DT model than in other models.