The Comparison of Two Method Nonparametric Approach on Small Area Estimation (Case: Approach with Kernel Methods and Local Polynomial Regression)

Small Area estimation is a technique used to estimate parameters of subpopulations with small sample sizes. Small area estimation is needed in obtaining information on a small area, such as sub-district or village. Generally, in some cases, small area estimation uses parametric modeling. But in fact, a lot of models have no linear relationship between the small area average and the covariate. This problem requires a non-parametric approach to solve, such as Kernel approach and Local Polynomial Regression (LPR). The purpose of this study is comparing the results of small area estimation using Kernel approach and LPR. Data used in this study are generated by simulation results using R language. Simulation data obtained by generating function m (x) are linear and quadratic pattern. The criteria used to compare the results of the simulation are Absolute Relative Bias (ARB), Mean Square Error (MSE), Generalized Cross Validation (GCV), and risk factors. The simulation results showed: 1) Kernel gives smaller relative bias than LPR does on both linear and quadratic data pattern. The relative bias obtained by Kernel tends to be more stable and consistent than the relative bias resulted by LPR, (2) the Kernel MSE is smaller than the LPR MSE either on linear or quadratic pattern in any combination treatment, (3) the value of GCV and the risk factors in Kernel are smaller than these in LPR in any combination of the simulated data patterns, (4) on non parametric data, for both linear data pattern and quadratic data pattern, Kernel methods provide better estimation compared to LPR.