Prediksi Kerawanan Wilayah Terhadap Tindak Pencurian Sepeda Motor Menggunakan Metode (S)ARIMA Dan CART

Abstract

Motor vehicle theft is a crime that is most common in Indonesia. Growth of vehicle motorcycle significant in each year accompanied by the increasing theft of motorcycles in each year, we need a system that is able to forecast the development and the theft of the motorcycle.

This research proposes the development of forecasting models vulnerability criminal offense of theft of motorcycles with ARIMA forecasting method. This method not only forecast from variable of theft but also residents, vehicles and unemployment. The study also determined the classification level of vulnerability to the crime of theft of a motorcycle using a method based on the Decision Tree CART ARIMA forecasting method.

Forecasting time series data with ARIMA method performed by each of the variables to produce the best ARIMA forecasting model which varies based on the data pattern of each of those variables. The results of classification by CART method to get the value of accuracy of 92% for the city of Yogyakarta and 85% for DIY. Based on the above, the results of ARIMA forecasting and classification CART can be used in determining the level of vulnerability to the crime of theft of motorcycles.