A Comparative Study of Machine Learning Models for COVID-19 prediction in India
Vartika Bhadana, Anand Singh Jalal, Pooja Pathak
Abstract
Machine learning is commonly being used in every field. Forecasting systems based on machine learning (ML) have shown their importance in interpreting perioperative effects to accelerate decision-making on the potential course of action. In several technology domains, ML models have been used long to define and prioritize adverse threat variables. To manage forecasting challenges, many prediction approaches are widely used. The paper shows the ability of ML models to estimate the amount of forthcoming COVID-19-affected patients that is now considered a serious threat to civilization. In paper this, we have performed a comparative study of five machine learning standard models like Linear regression (LR), decision tree, least absolute shrinkage and selection operator (LASSO), random forest and support vector machine (SVM) to forecast the threatening variables of COVID-19. Each of the models makes three forms of forecasts, i.e. the total active cases, the total deaths, and the total recoveries in the next five days. The findings provided by the paper suggest that the use of these techniques for the current COVID-19 the pandemic scenario is a promising strategy. For better accuracy, we have used a six-degree polynomial. Experiment results illustrate that poly LR and poly LASSO gives the best results followed by LR, LASSO, random forest, and decision tree. SVM shows the poor result in the prediction of COVID-19.