Using Criminal Statistics to Predict Crime Values through Time Series

Abstract

Objectives: The primary objective of this study was to ascertain the overarching pattern of criminal activity in the governorate of Sousse (Tunisia), spanning eleven years from January 2011 to December 2021. Additionally, the study aimed to forecast crime rates for the subsequent thirty-six months, providing insights to anticipate and elucidate future trends. The study's focal challenge was to leverage criminal statistics in constructing a robust statistical model for discerning the general trajectory of crime in Sousse and predicting its prospective trends.

Methods: This research adopted a descriptive-analytical approach and employed the Statistical Package for the Social Sciences (SPSS V.21) and Eviews V.13 programs to analyze the time series data of total crimes (comprising 155,829 incidents recorded from 2011 to 2021, with an annual average of 14,166 crimes) in the governorate. The analysis employed the Box-Jenkins method to establish a fitting model, subsequently utilizing this model for forecasting future values.

Results: Findings indicated a consistent upward trend in overall crime rates in Sousse. The time series analysis culminated in identifying the optimal model denoted by SARIMA (3,1,2) (0,1,1)₁₂, employed for forecasting total crime until the end of 2024. The projected series' overall trend component exhibited a decline in crime values over time, a phenomenon attributed to the political and security stability of the Tunisian state.

Conclusions: This study underscored the efficacy of the Box-Jenkins method in time series analysis as a premier and accurate statistical research approach for predicting crime and discerning its trends. The paper concluded with recommendations for security decision-makers to fortify security policies and plans. Further research is necessary to explore the application of time series as a tool for studying and predicting crime trends in Arab countries.