Predicción del número de delitos de tránsito que ingresaran a la fiscalía provincial de Chimborazo mediante ARIMA 2018-2021

Abstract

This research work predicts the number of traffic offenses that will enter the Chimborazo Provincial Prosecutor's Office (FPCH), using seasonal ARIMA models (SARIMA) so that with this knowledge it is possible to make more accurate decisions, improve the care and distribution of goods in the FPCH. To carry out the work, the monthly data for the 2015-2020 period provided by the Procedural Management unit of the Chimborazo Provincial Prosecutor's Office was used. Through a statistical analysis with the 80-20 rule, it was obtained that the most representative variable is injuries caused by traffic accidents (LCPATR) since it represents 83% of all traffic crimes, it is also taken into account the variable total traffic offenses (DTTR). Through an exploratory analysis of the data, the presence of atypical data was found in the two variables, in addition, the decomposition of the time series was made and it was observed that there is seasonality. Using the Box Jenkins methodology, the best models for the two series were found and, for the validation of the proposed models, the assumptions of normality, homoscedasticity and independence to errors were tested. For the LCPATR variable, a SARIMA (0,1,1) (0,1,1) [12] is presented as the best model, which complies with the validation assumptions, and for the DTTR variable, a SARIMA (0,1,1) (0,1,1) [12], which also meets the validation assumptions. Finally, predictions were obtained using the proposed models and it is appreciated that there will be an increase in the number of offenses by LCPATR in February and in the following months they will decrease. Similarly, with the DTTR variable, the predictions indicate that there will be an increase in the month of February and in the following months the number of offenses will decrease.