Penerapan Metode Least Absolute Shrinkage and Selection Operator (LASSO) untuk Mengatasi Multikolinearitas pada Data Spasial (Studi Kasus : Tingkat Kemiskinan di Sumatera Utara Tahun 2022)

Abstract

Spatial data is data that contains the influence of location with variance that is not homogeneous at each location or there is spatial heterogeneity. To overcome spatial heterogeneity, the Geographically Weighed Regression (GWR) model is used. However, in the GWR model there are symptoms of multicollinearity, namely a strong relationship between independent variables which will reduce the accuracy of parameter estimates. To overcome multicollinearity in the GWR model, the Least Absolute Shrinkage and Selection Operator (LASSO) method is used. The LASSO method estimates the parameters of the GWR model by minimizing the sum of the squared errors of the constraint function which is solved using the Least Angle Regression (LARS) algorithm. So we get the Least Absolute Shrinkage and Selection Operator (LASSO) regression model to overcome the problem of multicollinearity in spatial data. Based on the research results obtained, the LASSO method can overcome multicollinearity in the GWR model so that 33 final models are obtained. One model is Nias Regency with factors that influence the level of poverty in Nias Regency are open unemployment rate, life expectancy rate, average years of schooling, gross enrollment rate, and Per capita income.