Developing a predictive model for COVID-19 in Uganda

Stuart Muwanga Sebiranda; Peter Busch; Stephen Smith

Developing a predictive model for COVID-19 in Uganda

Stuart Muwanga Sebiranda^*, Peter Busch, Stephen Smith

^*Corresponding author for this work

School of Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

Abstract

While machine learning (ML) has been used in the fight against COVID-19, there is limited research using ML to make predictions and identify essential variables in a developing country in East Africa. With COVID-19 widening the gap between developing and developed countries, it is necessary to reduce this disparity, which is also noted as a significant UN Sustainable Development Goal. This paper aims to use ML models to compare the spread of COVID-19 in Uganda. The daily and cumulative number of COVID-19 cases are modelled using two machine learning models: linear regression and the cat boost regressor model. The cat boost model was the best-performing model with promising results. The significant variables in the cat boost model were then determined using SHAP values, which showed the number of government hospitals, GDP per capita and temperature, significantly impacted COVID-19 cases in Uganda

Original language	English
Title of host publication	Australasian Conference on Information Systems (ACIS)
Place of Publication	University of Canberra
Number of pages	11
Publication status	Submitted - 19 Aug 2024
Event	Australasian Conference on Information Systems 2024 : Digital Futures for a Sustainable Society - University of Canberra, Canberra, Australia Duration: 4 Dec 2024 → 6 Dec 2024 Conference number: 35th https://acis.aaisnet.org/acis2024/

Conference

Conference	Australasian Conference on Information Systems 2024
Abbreviated title	ACIS 2024
Country/Territory	Australia
City	Canberra
Period	4/12/24 → 6/12/24
Internet address	https://acis.aaisnet.org/acis2024/

Keywords

COVID-19
SHAP
Uganda
Machine learning (ML)
Cat Boost Regressor
Linear Regression

Cite this

@inproceedings{59ba81e5024c405dbd6e297f10faf4e1,

title = "Developing a predictive model for COVID-19 in Uganda",

abstract = "While machine learning (ML) has been used in the fight against COVID-19, there is limited research using ML to make predictions and identify essential variables in a developing country in East Africa. With COVID-19 widening the gap between developing and developed countries, it is necessary to reduce this disparity, which is also noted as a significant UN Sustainable Development Goal. This paper aims to use ML models to compare the spread of COVID-19 in Uganda. The daily and cumulative number of COVID-19 cases are modelled using two machine learning models: linear regression and the cat boost regressor model. The cat boost model was the best-performing model with promising results. The significant variables in the cat boost model were then determined using SHAP values, which showed the number of government hospitals, GDP per capita and temperature, significantly impacted COVID-19 cases in Uganda",

keywords = "COVID-19, SHAP, Uganda, Machine learning (ML), Cat Boost Regressor, Linear Regression",

author = "Sebiranda, {Stuart Muwanga} and Peter Busch and Stephen Smith",

year = "2024",

month = aug,

day = "19",

language = "English",

booktitle = "Australasian Conference on Information Systems (ACIS)",

note = "Australasian Conference on Information Systems 2024<br/> : Digital Futures for a Sustainable Society, ACIS 2024 ; Conference date: 04-12-2024 Through 06-12-2024",

url = "https://acis.aaisnet.org/acis2024/",

}

TY - GEN

T1 - Developing a predictive model for COVID-19 in Uganda

AU - Sebiranda, Stuart Muwanga

AU - Busch, Peter

AU - Smith, Stephen

N1 - Conference code: 35th

PY - 2024/8/19

Y1 - 2024/8/19

N2 - While machine learning (ML) has been used in the fight against COVID-19, there is limited research using ML to make predictions and identify essential variables in a developing country in East Africa. With COVID-19 widening the gap between developing and developed countries, it is necessary to reduce this disparity, which is also noted as a significant UN Sustainable Development Goal. This paper aims to use ML models to compare the spread of COVID-19 in Uganda. The daily and cumulative number of COVID-19 cases are modelled using two machine learning models: linear regression and the cat boost regressor model. The cat boost model was the best-performing model with promising results. The significant variables in the cat boost model were then determined using SHAP values, which showed the number of government hospitals, GDP per capita and temperature, significantly impacted COVID-19 cases in Uganda

AB - While machine learning (ML) has been used in the fight against COVID-19, there is limited research using ML to make predictions and identify essential variables in a developing country in East Africa. With COVID-19 widening the gap between developing and developed countries, it is necessary to reduce this disparity, which is also noted as a significant UN Sustainable Development Goal. This paper aims to use ML models to compare the spread of COVID-19 in Uganda. The daily and cumulative number of COVID-19 cases are modelled using two machine learning models: linear regression and the cat boost regressor model. The cat boost model was the best-performing model with promising results. The significant variables in the cat boost model were then determined using SHAP values, which showed the number of government hospitals, GDP per capita and temperature, significantly impacted COVID-19 cases in Uganda

KW - COVID-19

KW - SHAP

KW - Uganda

KW - Machine learning (ML)

KW - Cat Boost Regressor

KW - Linear Regression

M3 - Conference proceeding contribution

BT - Australasian Conference on Information Systems (ACIS)

CY - University of Canberra

T2 - Australasian Conference on Information Systems 2024<br/>

Y2 - 4 December 2024 through 6 December 2024

ER -