Estimasi Parameter Model Regresi Nonparametrik Birespon berdasarkan Penalized Spline Pada Data Tindak Kriminal di Indonesia (Studi Kasus Jumlah Kejadian Kejahatan terhadap Kesusilaan dan Jumlah Kejadian Kejahatan terhadap Fisik di Indonesia Tahun 2020)

Abstract

To achieve the creation of a safe and peaceful social life, crime is one of the things that is very concerned. In 2020, in Indonesia there were 6.872 crimes against decency and 36.672 crimes against the physical. One of the efforts that can be done to reduce the number of incidents of crimes against decency and the number of incidents of physical crimes in Indonesia is to model it on the factors that influence it so that predictions can be obtained. In this study, an estimation of the parameters of the nonparametric biresponse regression model was carried out based on the penalized spline estimator using the Weighted Least Square (WLS) method approach to predict the number of crimes against decency and the number of incidents of physical crime in Indonesia in 2020 with the predictor variable population density (X 1 ), the ratio of sex gender (X 2 ), the percentage of poor people (X 3 ) and the average net wage of workers/employees/employees (X 4 ). The penalized spline estimator is used to calculate knot points and smoothing parameters simultaneously so as to produce accuracy and smoothness of the curve shape simultaneously. The best model depends on determining the knot point and optimal smoothing parameters, namely the minimum Generalized Cross Validation (GCV) value. The best model is obtained when the number of knots for X 1 is one, X 2 is three, X 3 is three, and X 4 is one and λ = 0, 000000171 with a GCV of 568.359 and coefficient of determination of 0.652.Keywords: the number of incidents of crimes against decency, the number of incidents of crimes against physical, biresponse nonparametric regression, penalized spline, WLS, GCV.