The Accuracy of Various Forecasting Models on Creating Weekly Predictions on the Number of COVID-19 Cases

School Name

Spring Valley High School

Grade Level

10th Grade

Presentation Topic

Computer Science

Presentation Type

Non-Mentored

Abstract

COVID-19 creates an overwhelming influx of patients that hospitals could better prepare for if they could accurately predict the number of COVID-19 cases. The purpose of this study was to identify an accurate forecasting model that hospitals could use to predict the number of COVID-19 cases. It was hypothesized that if the Single Exponential Smoothing (SES), Autoregressive Integrated Moving Average (ARIMA), and Seasonal Autoregressive Integrated Moving Average (SARIMA) models were individually used and combined to predict the weekly number of COVID-19 cases, then the model combining the SES and SARIMA models would be the most accurate since it accounts for multiple irregularities. Using Python, the SES, ARIMA, and SARIMA models were generated from the weekly number of COVID-19 cases in the United States from January 29, 2020, to December 30, 2021, and were used to forecast the weekly number of COVID-19 cases from January 5, 2022, to November 30, 2022. The combinatorial models summed the results of the two combined models after multiplying the results by a value representing that model’s error. The SES model was the most consistent, with the lowest SD, and most accurate, with the lowest mean absolute percent error. The results of a two-way ANOVA test without replication (�� = .05) suggested that the data were significant [F(4, 188) = 2.513, p = .043], and there were notable differences in the forecasting models’ results. As such, the SES model would be the best forecasting system to be established in hospitals for better resource allocation and patient treatment.

Location

ECL 105

Start Date

3-25-2023 11:00 AM

Presentation Format

Oral and Written

Group Project

No

COinS
 
Mar 25th, 11:00 AM

The Accuracy of Various Forecasting Models on Creating Weekly Predictions on the Number of COVID-19 Cases

ECL 105

COVID-19 creates an overwhelming influx of patients that hospitals could better prepare for if they could accurately predict the number of COVID-19 cases. The purpose of this study was to identify an accurate forecasting model that hospitals could use to predict the number of COVID-19 cases. It was hypothesized that if the Single Exponential Smoothing (SES), Autoregressive Integrated Moving Average (ARIMA), and Seasonal Autoregressive Integrated Moving Average (SARIMA) models were individually used and combined to predict the weekly number of COVID-19 cases, then the model combining the SES and SARIMA models would be the most accurate since it accounts for multiple irregularities. Using Python, the SES, ARIMA, and SARIMA models were generated from the weekly number of COVID-19 cases in the United States from January 29, 2020, to December 30, 2021, and were used to forecast the weekly number of COVID-19 cases from January 5, 2022, to November 30, 2022. The combinatorial models summed the results of the two combined models after multiplying the results by a value representing that model’s error. The SES model was the most consistent, with the lowest SD, and most accurate, with the lowest mean absolute percent error. The results of a two-way ANOVA test without replication (�� = .05) suggested that the data were significant [F(4, 188) = 2.513, p = .043], and there were notable differences in the forecasting models’ results. As such, the SES model would be the best forecasting system to be established in hospitals for better resource allocation and patient treatment.