Modelización estadística - informática con r de delitos en la provincia de Chimborazo

Abstract

The present research work proposed to implement a SARIMA model in order to make a forecast of the number of crimes in the province of Chimborazo. For the execution of the investigation, the information through monthly crime data that concerns the period January 2015-April 2022 provided by the Provincial Prosecutor's Office of Chimborazo was used. For the modeling process, the Box-Jenkins methodology was utilized employing the R statistical software. Through univariate exploratory analysis, the presence of anomalous data was detected, which was replaced with the average number of crimes corresponding to the month of the atypical data. Once the series was stationary, with an autocorrelation analysis, the AR and MA parameters of the regular and seasonal part were estimated. Applying the auto.arima function of R and the proposed models, the SARIMA(2,1,1)(1,0,0)[12] model was obtained, being the most appropriate. Once the best model was determined, it was verified that it complies with the assumptions of normality, independence and homoscedasticity, thus guaranteeing that the model adequately adjusts to the data. Finally, the forecast of crimes for the next 12 months in the period May 2022-April 2023 was done, indicating that there will be an increase in the month of May with a total of 727 crimes, this being the highest compared to the remaining months that descend ranging from 680 to 696 monthly crimes, so it is recommended that the Provincial Prosecutor’s Office of Chimborazo take control actions and strengthen security during those months.