Efficient time series of smoothing and auto-regressive forecasting models for predicting police officer fatalities in the USA

Nagappan, D, Jayabalan, M ORCID: 0000-0002-1599-965X, Alanezi, A, Nadi, F and Coombs, T (2024) 'Efficient time series of smoothing and auto-regressive forecasting models for predicting police officer fatalities in the USA.' In: Bee Wah, Y, Al-Jumeily OBE, D and Berry, M.W, eds. Data science and emerging technologies. DaSET 2023. Springer, Singapore, pp. 181-192. ISBN 9789819702923

Abstract

With police deaths rising, predicting the number of police deaths is now of significant importance and is necessary to take precautions to prevent deaths from affecting the police force, public, and government from an associated reduction in police numbers. The aim of this research is to compare different time series forecasting models and find the most efficient model in predicting police deaths occurring in the USA. The dataset used in this study consisted of details of police officers who had died in service on duty in the USA. A total of 26,269 records between January 3, 1791, and June 3, 2022. The dataset was obtained through the Kaggle website data repository. A total of four smoothing models and four auto-regressive models were used and compared in this research. The smoothing models had better RMSE and MAPE scores, with HWES being the best-performing model. In summary, the HWES model performed the best on the USA police deaths dataset by producing less error compared to the other smoothing and auto-regressive models used in this research.

Item Type:	Book Chapter or Section
Divisions:	Bath School of Design
Date Deposited:	22 Aug 2025 15:56
Last Modified:	22 Aug 2025 15:56
URN:	https://researchspace.bathspa.ac.uk/id/eprint/17223

	Request a change to this item or report an issue
	Update item (repository staff only)