PEMODELAN STATUS KESEJAHTERAAN DAERAH KABUPATEN ATAU KOTA DI JAWA TENGAH MENGGUNAKAN GEOGRAPHICALLY WEIGHTED LOGISTIC REGRESSION SEMIPARAMETRIC

Abstract

Welfare in society is one of the most important aspects in ensuring the realization of the social where people have a good level of welfare. Benchmarks achieved prosperity is the fulfillment of basic needs of society as feasible. Statistical methods have been developed for the analysis of spatial data by taking into account factors that Geographically Weighted Logistic Regression Semiparametric (GWLRS). GWLRS is a local form of the logistic regression where there are parameters that are influenced by the location (Geographically varying coefficient) and the parameters that are not influenced by the location (fixed coefficient). Selection of the optimum bandwidth using Cross Validation (CV). Model GWLRS Welfare Status district or city in Central Java showed that GWLRS models differ significantly from the logistic regression model. And models generated for each area will be different from each other. To get the best models, the number of models to be evaluated. One method for selecting the best model is the value of the Akaike Information Criterion (AIC). Based on AIC obtained the best model is the model GWLRS because it has the smallest AIC value of 46.11213 with a classification accuracy of 77.14%. Keywords: Welfare, Geographically Weighted Logistic Regression Semiparametric, Cross Validation, Akaike Information Criterion