tetano
Editor, Senior Moderator
Comput Struct Biotechnol J
. 2020 Sep 24.
doi: 10.1016/j.csbj.2020.09.015. Online ahead of print.
Tailoring Time Series Models For Forecasting Coronavirus Spread: Case Studies of 187 Countries
Leila Ismail[SUP] 1 [/SUP], Huned Materwala[SUP] 1 [/SUP], Taieb Znati[SUP] 2 [/SUP], Sherzod Turaev[SUP] 1 [/SUP], Moien A B Khan[SUP] 3 4 [/SUP]
Affiliations
Abstract
When will the coronavirus end? Are the current precautionary measures effective? To answer these questions it is important to forecast regularly and accurately the spread of COVID-19 infections. Different time series forecasting models have been applied in the literature to tackle the pandemic situation. The current research efforts developed few of these models and validates its accuracy for selected countries. It becomes difficult to draw an objective comparison between the performance of these models at a global scale. This is because, the time series trend for the infection differs between the countries depending on the strategies adopted by the healthcare organizations to decrease the spread. Consequently, it is important to develop a tailored model for a country that allows healthcare organizations to better judge the effect of the undertaken precautionary measures, and provision more efficiently the needed resources to face this disease. This paper addresses this void. We develop and compare the performance of the time series models in the literature in terms of root mean squared error and mean absolute percentage error.
Keywords: COVID-19; Coronavirus; Epidemic transmission; Forecasting models; Machine learning models; Pandemic; Time series models.
. 2020 Sep 24.
doi: 10.1016/j.csbj.2020.09.015. Online ahead of print.
Tailoring Time Series Models For Forecasting Coronavirus Spread: Case Studies of 187 Countries
Leila Ismail[SUP] 1 [/SUP], Huned Materwala[SUP] 1 [/SUP], Taieb Znati[SUP] 2 [/SUP], Sherzod Turaev[SUP] 1 [/SUP], Moien A B Khan[SUP] 3 4 [/SUP]
Affiliations
- PMID: 32994886
- PMCID: PMC7513749
- DOI: 10.1016/j.csbj.2020.09.015
Abstract
When will the coronavirus end? Are the current precautionary measures effective? To answer these questions it is important to forecast regularly and accurately the spread of COVID-19 infections. Different time series forecasting models have been applied in the literature to tackle the pandemic situation. The current research efforts developed few of these models and validates its accuracy for selected countries. It becomes difficult to draw an objective comparison between the performance of these models at a global scale. This is because, the time series trend for the infection differs between the countries depending on the strategies adopted by the healthcare organizations to decrease the spread. Consequently, it is important to develop a tailored model for a country that allows healthcare organizations to better judge the effect of the undertaken precautionary measures, and provision more efficiently the needed resources to face this disease. This paper addresses this void. We develop and compare the performance of the time series models in the literature in terms of root mean squared error and mean absolute percentage error.
Keywords: COVID-19; Coronavirus; Epidemic transmission; Forecasting models; Machine learning models; Pandemic; Time series models.