Pronóstico en series de tiempo de los hurtos a personas en Bogotá para 2024.
Cargando...
Fecha
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
Contexto: Entre 2018 y 2023, Bogotá vio un incremento sostenido en los hurtos a personas, alcanzando más de 147,000 incidentes en 2023. Este aumento refleja una creciente preocupación por la seguridad ciudadana. La evaluación de modelos de series de tiempo es clave para analizar tendencias y predecir incidentes futuros, ayudando a diseñar estrategias preventivas efectivas. Propósito: Generar conocimiento en ciencia de datos y estadística aplicada con base en un análisis de serie temporal de hurtos en Bogotá mediante técnicas avanzadas de estadística para identificar sus principales componentes y proporcionar una modelación robusta de sus patrones. Metodología: Se recolecto los datos públicos del a Secretaría de Seguridad, Convivencia y Justicia de Bogotá, y con base en la metodologia Box-Jenkins fue posible el modelaje SARIMA para el pronóstico de hurtos en la ciudad. Resultados: Se determino que un modelo SARIMA (0,1,1)(1,0,0) el cual tiene un equilibiro optimo con un AIC de 1215.06 y un error promedio de 1183.63 puntos (hurtos) en su predicción, adicionalmente hace un pronostico estable y en concordacia con los datos del año 2024. El modelo presento un ajuste cercano a la distribución normal, con facultades de homocedasticidad y sin auto correlaciones en sus residuos. Conclusiones: El modelo SARIMA propuesto genera un pronostico estable, y permite entender que la cantidad de hurtos en la ciudad, como reflejo de la creciente inseguridad se ve afectado por factores políticos, demográficos y que los datos futuros depende en buena media de sus periodos inmediatamente anteriores.
Context: Between 2018 and 2023, Bogota saw a sustained increase in thefts from persons, reaching more than 147,000 incidents in 2023. This increase reflects a growing concern for citizen security. The evaluation of time series models is key to analyze trends and predict future incidents, helping to design effective preventive strategies. Purpose: Generate knowledge in data science and applied statistics based on a time series analysis of theft in Bogota using advanced statistical techniques to identify its main components and provide a robust modeling of its patterns. Methodology: Public data was collected from the Secretariat of Security, Coexistence and Justice of Bogota, and based on the Box-Jenkins methodology, SARIMA modeling was possible to forecast theft in the city. Results: It was determined that a SARIMA (0,1,1)(1,0,0) model, which has an optimal equilibrium with an AIC of 1215.06 and an average error of 1183.63 points (thefts) in its prediction, additionally makes a stable forecast and in agreement with the data for the year 2024. The model presented a fit close to the normal distribution, with homoscedasticity and no autocorrelations in its residuals. Conclusions: The proposed SARIMA model generates a stable forecast, and allows us to understand that the number of thefts in the city, as a reflection of the growing insecurity, is affected by political and demographic factors, and that future data depends on a good average of its immediately preceding periods.
Context: Between 2018 and 2023, Bogota saw a sustained increase in thefts from persons, reaching more than 147,000 incidents in 2023. This increase reflects a growing concern for citizen security. The evaluation of time series models is key to analyze trends and predict future incidents, helping to design effective preventive strategies. Purpose: Generate knowledge in data science and applied statistics based on a time series analysis of theft in Bogota using advanced statistical techniques to identify its main components and provide a robust modeling of its patterns. Methodology: Public data was collected from the Secretariat of Security, Coexistence and Justice of Bogota, and based on the Box-Jenkins methodology, SARIMA modeling was possible to forecast theft in the city. Results: It was determined that a SARIMA (0,1,1)(1,0,0) model, which has an optimal equilibrium with an AIC of 1215.06 and an average error of 1183.63 points (thefts) in its prediction, additionally makes a stable forecast and in agreement with the data for the year 2024. The model presented a fit close to the normal distribution, with homoscedasticity and no autocorrelations in its residuals. Conclusions: The proposed SARIMA model generates a stable forecast, and allows us to understand that the number of thefts in the city, as a reflection of the growing insecurity, is affected by political and demographic factors, and that future data depends on a good average of its immediately preceding periods.