Multivariate Forecasting of Homicide Count for the Metropolitan Zone of Guadalajara, México

The Metropolitan Area of Guadalajara (MAG) is the result of an urban conurbation phenomenon involving the capital of the State of Jalisco. It stands as the most politically, economically, and socially significant region in the central part of Jalisco. However, this prominence also makes it a highly criminal area, with high homicide rates driven by various social phenomena, including organized crime and other social factors. One of the most crucial challenges for governance in Jalisco or any of the municipalities that comprises the MAG is the coordination of security forces to effectively reduce homicides. In 2023, the homicide rate decreased compared to the previous year, a figure that could be attributed to the efforts within the MAG to coordinate municipal police forces. However, this also suggests that crimes in neighboring municipalities may help predicting crimes in a MAG municipality. In this study, we conducted a comparison of the SARIMAX forecasting model, using as exogenous variables other crimes committed in the neighboring municipalities that also constitute the MAG, to improve shortterm (1 and 2 months) prediction of the Guadalajara homicide count. In addition to the improevment of the forecasting results, identifying the features that best predict future crimes sheds light on the relation among different types of crime and neighboring geographical demarcations.