The theft criminality forecasting in the surabaya district police region using poisson GARMA model and negative binomial GARMA model. (March 2020)

• Record Type: Journal Article
• Authors: Wardhani, L P; Indrawati, D; Wahyuningsih, N; Setiawan, ; Suhartono, ; Kuswanto, H
• Is Part Of: Journal of physics. Volume 1490(2020)
• Publication Date: 2020-03
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1490/1/012015
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: Theft is a crime that often occurs in the community. According to the Police, theft consists of ordinary theft, petty theft, weight theft, and theft of violence. According to BPS, the highest number of incidents of theft occurred during 2018, one of which was in the Surabaya City Police Region. Based on these facts forecasting is done to provide material for consideration of the number of incidents of theft in the future. Forecasting generally uses the ARMA model. The ARMA model is less suitable for use in count data. The ARMA model was developed into the GARMA model if the data is assumed to be a Poisson distribution called the Poisson GARMA model. If the data is assumed to be a Negative Binomial distribution is called the Negative Binomial GARMA model. The estimation of the Poisson GARMA model and the Binomial Negative GARMA model uses MLE and is optimized using IRLS. This model is applied in cases of theft crimes in the Surabaya City Police Region and compares the accuracy of forecasting between the Poisson GARMA model and the Negative Binomial GARMA model. Based on the analysis carried out, the Negative Binomial GARMA (1.0) model is better for Central Surabaya, East Surabaya, and South Surabaya than Poisson GARMA (1.0) model. The Negative Binomial GARMA (0.2) model is better for the West Surabaya region rather than the Poisson GARMA (0.2) model. The best model in the Surabaya City Police Region area is the Binomial Negative GARMA (0.2) model in the West Surabaya region. The selection of the best model uses the lowest value of AIC.