Modeling Industrial and Maritime Sector Effects on Economic Growth in Riau Island Using GWNR-P Spline

Abstract

Economic growth is an important indicator of regional development and is influenced by various socioeconomic factors. In the Riau Islands Province, the effects of the Labor Force Participation Rate (LFPR) and Capture Fishery Production Value on the Gross Regional Domestic Product (GRDP) percentage distribution may vary across districts and cities due to spatial heterogeneity and nonlinear relationships. Therefore, this study applies the GWNR-PSpline model to analyze these relationships. The optimal Penalized Spline model was obtained using knot points of 67.83 and 15.67 with a smoothing parameter of 50, as indicated by the minimum Generalized Cross Validation (GCV) value of 1245.917. Subsequently, the Manhattan distance and Bisquare kernel weighting scheme produced the minimum Cross Validation (CV) value of 64.51, indicating the lowest prediction error among the evaluated weighting combinations. The estimated local parameters varied across districts and cities, indicating spatial heterogeneity in the relationships between the explanatory variables and economic growth. Spatial analysis revealed that Natuna and Anambas Island exhibited stronger local effects than other regions. Furthermore, the local coefficients of determination ranged from 0.3944 to 0.8603, substantially exceeding the global P-Spline model (R2=0.158). These findings demonstrate that the GWNR-PSpline model effectively captures both nonlinear and spatially varying relationships.