Covid-19 forecasting and analysis using different time-series model and algorithms

Abstract

Introduction: The novel coronavirus (COVID-19) has significantly spread over the world and impacted with new challenges to the research community. Although governments initiated numerous containment and social distancing measures all over the world, the need for healthcare resources has dramatically increased and the effective management of infected patients becomes a challenging task for healthcare centres. Objective: Thus, the objective of the research is to find the accurate short-term forecasting of the number of new confirmed covid-19 positive cases is important for optimizing the available resources and slowing down the progression of COVID-19. Recently, various methods like machine learning models and other algorithms demonstrated important improvements when handling time-series data in various applications. Methods: This paper presents a comparative study of different machine learning methods and models to forecast the number of new cases. Specifically, Long short-term memory (LSTM), Autoregressive Integrated Moving Average (ARIMA), Holt’s Linear forecasting model, Exponential smoothing and Moving-average model algorithms have been applied for forecasting of COVID-19 cases based on data set. Result: Results were analysed using various parameters like Root Mean Square Error, Mean Absolute Error, Mean Absolute Percentage Error, Error Vector Magnitude Root Mean Square Logarithmic Error. Conclusion: As a conclusion, compared to other models, Long Short TermModel predicted better forecasting and gives the best performance in terms of different parameters.