Multiple Models for Forecasting Case Rate and Death Rate of COVID-19

Public Deposited

Analytics

Last Modified

Creator

Li, Rui
- Affiliation: College of Arts and Sciences, Department of Statistics and Operations Research

Abstract

Within the last two decades, coronaviruses have generated devastating effects on humans being. Especially in 2020, the appearance of new coronaviruses, COVID-19, makes researchers aware of the importance of forecasting and predicting the COIVD-19 spreading patterns. This research utilizes four different time series models - Naïve model, ARIMA model, Horizon-Specific Model without Difference, and Horizon-Specific Model with Difference – to predict the case and death rate with a horizon time of 1 day, 3 days, and 5 days ahead for 173 different countries. We identify the best model for each country based on the minimization of out-of-sample root mean squared error (RMSE). The results show that when forecasting the case rate, ARIMA models are the best fit for around 54% of countries with a horizon time of 1 day ahead, while the horizon-specific models without difference are the most appropriate for about 50% and 49% of countries with a horizon time of 3 days and 5 days ahead, respectively. When forecasting the death rate, the ARIMA models significantly outperform the others and are suitable for 59%, 39%, and 40% of countries with a horizon of 1 day, 3 days, and 5 days ahead, respectively.

Date of publication

DOI

Resource type

Rights statement

Advisor

Giacomazzo, Mario
- Affiliation: College of Arts and Sciences, Department of Statistics and Operations Research

Degree

Graduation year

Language

Relations

Parents:

This work has no parents.

In Collection:

Thumbnail	Title	Date Uploaded	Visibility	Actions
	Rui_Li-Senior_Honor_Thesis_Project.pdf	2021-05-02	Public	Download