Leveraging Ensemble and Hybrid Forecasting Tools to Increase Accuracy: Turkey COVID-19 Case Study

Date

2025-02-01

Author

Evkaya, O. Ozan
KURNAZ, Fatma Sevinç
Özdemir, Özlem
YİĞİT, Pakize

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

23
views

0
downloads

Cite This

In order to cope with spread of global disease such as coronavirus (COVID-19), it is important to achieve a more accurate and efficient predictive model by leveraging the various methods for capturing both linear and non-linear patterns. Even if COVID-19 is ’no longer a global emergency’, still ranking it fifth in the deadliest epidemics historically. In that respect, it is crucial to derive plausible forecasts, and correspondingly various related papers on COVID-19 cases have been published during the period of 2020–2022. This article mainly aims to explore several time series forecasting methods to predict the spread of COVID-19 during the pandemic’s in Turkey over two different period of times. The COVID-19 data is retrieved from the website of the Turkish Ministry of Health over two different time periods based on the fact that how the number of cases are announced in Turkey (Period 1: March 2020-May 2022 and Period 2: November 2020-May 2022). In addition to classical techniques such as the AutoRegressive Moving Average (ARIMA) model, innovations state space models for exponential smoothing (ETS), the neural network autoregression (NNETAR) model, the novel ensemble and hybrid type of methods are studied to forecast the number of cases and deaths in Turkey. The main results showed that ensemble and hybrid models are better at capturing the nonlinear, and seasonal pandemic patterns, significantly outperforming the respective single models. These findings are consistent over different time periods so that for short-term forecasts, they can be more useful for decision-making.

Subject Keywords

COVID-19, Daily cases and deaths, Ensemble models, Hybrid models

URI

https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85218187399&origin=inward
https://hdl.handle.net/11511/113893

Collections

Department of Business Administration, Article