A Multi-Step Predictive Model for COVID-19 Cases in Nigeria Using Machine Learning

Sakinat Oluwabukonla Folorunso (), Ezekiel Adebayo Ogundepo (), Joseph Bamidele Awotunde (), Femi Emmanuel Ayo (), Oluwatobi Oluwaseyi Banjo () and Abass Ishola Taiwo ()
Additional contact information
Sakinat Oluwabukonla Folorunso: Olabisi Onabanjo University
Ezekiel Adebayo Ogundepo: Data Science Nigeria
Joseph Bamidele Awotunde: University of Ilorin
Femi Emmanuel Ayo: McPherson University
Oluwatobi Oluwaseyi Banjo: Olabisi Onabanjo University
Abass Ishola Taiwo: Olabisi Onabanjo University

Chapter Chapter 7 in Decision Sciences for COVID-19, 2022, pp 107-136 from Springer

Abstract: Abstract COronaVIrus Disease 2019 (COVID-19) is a vastly communicable disease that has the whole world in a panic state with a huge negative impact. Forecasting for future occurrences for timely intervention requires a good model. The aim of this chapter is to model a multistep forecast for 14-days (2 weeks) incidence of COVID-19 daily cases in Nigeria with Machine Learning (ML) models. The study dataset contains 241 instances (days) of the daily incidence of COVID-19 for confirmed, recovered, and patients who eventually died from February 27th to October 24th, 2020 in Nigeria. The proposed methodology jointly model multiple targets field (cases) simultaneously in order to capture dependencies between them. This 3-daily time-series study dataset was transformed to a supervised learning format for analysis by ML models. Linear Regression (LR), MultiLayerPerceptron Regressor (MLPR), Support Vector Regressor (SVR), k-Nearest Neighbor Regressor (k-NNR), and Random Forest Regressor (RFR) models were employed and their prediction errors were compared based on average values of Directional Accuracy (DAC), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and Root Mean Squared Error (RMSE). The result obtained showed that SVR outperformed all other models and can accurately forecast a 14-day occurrence of COVID-19 cases in Nigeria based on all the metrics used in this study. This study will aid the Government and health practitioners to plan ahead for the next 14 days (2 weeks). This 14-ahead step forecasting model could also be extended for longer terms. ML model-based forecasting can effectively model a small sample size and is useful for decision-making and planning for future pandemics by the government and health service workers.

Keywords: COVID-19; Regression; Machine learning; Time-series; Pandemic decision science (search for similar items in EconPapers)
Date: 2022
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:isochp:978-3-030-87019-5_7

Ordering information: This item can be ordered from
http://www.springer.com/9783030870195

DOI: 10.1007/978-3-030-87019-5_7

Access Statistics for this chapter

More chapters in International Series in Operations Research & Management Science from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().