ANALISIS SUMBER-SUMBER PENDAPATAN DAERAH KABUPATEN DAN KOTA DI JAWA TENGAH DENGAN METODE GEOGRAPHICALLY WEIGHTED PRINCIPAL COMPONENTS ANALYSIS (GWPCA)

Abstract

The districts/cities sources of revenue in Central Java consists of Natural Revenue District (PAD), the equalization fund (DAPER), and other local income. PAD consists of four variables namely local tax (X1) , retribution (X2) , the results of regional company and wealth management that is separated (X3) , and other legal PAD (X4). DAPER consists of four variables namely sharing of tax revenue (X5) , sharing of non-tax revenue (X6) , the general allocation fund (X7) , and the special allocation fund (X8). Other region revenues (X9) is a source of local income that is not included in the PAD or DAPER. Sources of local revenue variables are mutually correlated multivariate data and have spatial effect. Therefore Geographically Weighted Principal Components Analysis (GWPCA) is suitable for analyzing sources of local revenue variables. GWPCA is a multivariate analysis method that is used to eliminate multicolliniearity in the multivariate data that have spatial effect. The result of this study is that the variables of revenue sources on each location can be replaced by three new variables called PC1, PC2, and PC3 which is independent each other. Variance Cumulative Proportion that can be explained by those new variables is approximately 80%. Based on the first principal component (PC1) that have variance proportion approximately 50%, there are three groups which has different carracteristics. The first group is the region that the revenue have influenced by variables X9 followed by X1. The second group is the region that the revenue have influenced by variables X9 followed by X2. The third group is the region that the revenue have influenced by variables X9 followed by X5. It is also seen that Kudus District has the most distinct characteristics which the revenue are influenced by variables X5 followed by X9. Keywords : Source of Regional Revenue, Spatial Effect, Multivariate, Multicollinearity, GWPCA, Variance Proportion.