Nonparametric Analysis of Tracking Data in the Context of COVID-19 Pandemic

Thanks to the growth in data storage capacity, nowadays, researchers can use years’ worth of mathematical models and depend on past datasets. A pattern of all pandemics can be identified through the assistance of Machine Learning. The movement of the COVID-19 herd and any future pandemic can be predicted. These predictions will vary based on the dataset, but it will allow the preparation beforehand and stop the spreading of COVID-19. This study focuses on developing Spatio-temporal models using Machine Learning to produce a predictive visualized heat regional map of COVID-19 worldwide. Different models of Machine Learning are compared using John Hopkins University dataset. This study has compared well-known basic models like Support Vector Machine (SVM), Prophet, Bayesian Ridge Regression, and Polynomial Regression. Based on the comparison of various metrics of the Support Vector Machine, Polynomial Regression Model was found to be better and hence can be assumed to give good results for long-term prediction. On the other hand, ARIMA, Prophet Model, and Bayesian Ridge Reduction models are good for short-term predictions. The metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Square Error (RMSE) are better for Support Vector Machines compared to other models. The metrics such as R2 Score and Adjusted R-Square are better for the polynomial Regression model.

show abstract

Nonparametric Analysis of Tracking Data in the Context of COVID-19 Pandemic

Cited by 2 publications

References 27 publications

Review of Big Data Analytics and Machine Learning Models for Contract Tracking and Detecting of COVID-19 Pandemic Cases

Review of Big Data Analytics and Machine Learning Models for Contract Tracking and Detecting of COVID-19 Pandemic Cases

Forecasting Geo Location of COVID-19 Herd

Contact Info

Product

Resources

About