Machine Learning-based forecasting models for COVID-19 spread in Algeria
Authors:
- 23 May 2022
- journal article
- research article
- Published by SAGE Publications in Model Assisted Statistics and Applications
- Vol. 17 (2) , 99-108
Abstract
Currently, the Algerian health system is facing the fourth wave of COVID-19 in which the number of recovered cases grows exponentially each day due to the COVID-19 Omicron variant. According to the Algerian National Institute of Public Health (ANIPH), it was reported 168 668 confirmed cases and 4 189 deaths till 29 July, 2021. In this work, we aim to utilize supervised Machine Learning (ML) based models in an attempt to forecast the future trend of the disease in Algeria. To that end, we use three forecasting models: Facebook Prophet, LSTM and ARIMA. Forecasting results of the 90 future days are provided. The used dataset contains the confirmed and death cases collected from the daily Epidemiological Situation (ES), published by ANIPH, from 19 April 2020 to 29 July 2021. The forecasting accuracy of the models are assessed and compared using several statistical assessment criteria. The results show that ARIMA outperforms Facebook Prophet and LSTM in the case of confirmed cases. However, LSTM shows best performance in the case of death cases. This study shows clearly that the pandemic spread is still in progress and protection measures like contact restriction and lockdown should be strictly applied especially with the appearance of the COVID-19 Delta and Omicron variants.Keywords
This publication has 13 references indexed in Scilit:
- Prediction for the spread of COVID-19 in India and effectiveness of preventive measuresScience of The Total Environment, 2020
- Prediction of confinement effects on the number of Covid-19 outbreak in AlgeriaMathematical Modelling of Natural Phenomena, 2020
- Predictions for COVID-19 Outbreak in India using epidemiological modelsmedRxiv, 2020
- Predicting of COVID-19 Confirmed Cases in Different Countries with ARIMA Models in 2020medRxiv, 2020
- A Comparison of ARIMA and LSTM in Forecasting Time Series2018
- Forecasting at ScaleThe American Statistician, 2018
- Stan: A Probabilistic Programming Language2017
- Some Tests for Seasonality in Time Series DataJournal of Modern Applied Statistical Methods, 2016
- Long Short-Term MemoryNeural Computation, 1997
- Time Series Analysis: Forecasting and ControlJournal of Marketing Research, 1977