PEMILIHAN MODEL REGRESI NONPARAMETRIK TERBAIK UNTUK ANALISIS DATA INFLASI DI JAWA TENGAH

Abstract

Inflation becomes one of important problems as parameter of economic growth and determiner factor for government in formulating fiscal, monetary and nonmonetary policy. But, these days the policies were arranged can’t give the positive response to inflation pressure in the future. Therefore, the prediction of inflation rates are needed. Inflation rates are predicted by nonparametric regression approach because of the fluctuation of inflation which can’t be solved by classic time series models. In this research, the best nonparametric regression models are selected between kernel, local polynomial and spline regression to predict Central Java Inflation movement in 2014. Based on analysis, the best nonparametric regression is spline order 2, knot points are 5,37; 5,44; 5,59 and 9,01 with GCV 0,4367286. By using that model, the prediction of Central Java inflation got down since October 2013 until February 2014 on level 7% and March until December 2014 on level 6%. Keywords: inflation, kernel, local polynomial, spline