Time series analysis and prediction of COVID-19 patients using discrete wavelet transform and auto-regressive integrated moving average model.

Coronavirus disease (COVID-19) is a respiratory condition that has quickly expanded to pandemic levels, affecting people in over 192 countries around the world. Estimating the number of new COVID-19 cases is never an easy undertaking. This study explores the application of Auto-Regressive Integrated Moving Average (ARIMA) modelling combined with Discrete Wavelet Transform (DWT) for analyzing and predicting time series data of Covid-19 patients. It aims to comprehend the disease progression and forecast future trends. The methodology makes use of the ARIMA model, a statistical tool that captures temporal correlations, K-means clustering for outlier detection and removal, and the discrete wavelet transform (DWT) for feature selection. This combination allows for a more thorough evaluation of the patterns and deviations in the COVID-19 patient data. The Enhanced ARIMA (EARIMA) model aims to enhance the current ARIMA-based time series prediction methods. By employing this hybrid approach, the study aims to provide a robust framework for understanding the temporal dynamics of Covid-19 patient data, offering more accurate and interpretable predictions for future trends in the progression of the disease. It improves on the current ARIMA based time-series prediction algorithms. The EARIMA was evaluated using COVID-19 data set gotten from the WHO official website and was compared with various state-of-the-art algorithms of today. The EARIMA received an RSS score of 0.326 and an accuracy of 98.7%, which is better than the existing ARIMA model, which has an RSS score of 0.659 and an accuracy of 93%. The findings contribute to enhancing forecasting accuracy and aiding in decision-making processes for healthcare resource allocation and pandemic management strategies. [ABSTRACT FROM AUTHOR]