Dedeh Kurniasih, - and Scolastika Mariani, - and Sugiman, - (2013) Efisiensi Relatif Estimator Fungsi Kernel Gaussian Terhadap Estimator Polinomial Dalam Peramalan USD Terhadap JPY. Unnes Journal of Mathematics, 2 (2). pp. 79-84. ISSN 2460-5859

PDF - Published Version Download (750kB)

Abstract

The purpose of this paper is (1) analyze the relative efficiency from estimator Gaussian kernel function of the polynomial estimator, (2) compare the MSE of both estimators and (3) determine the exchange rate of USD against JPY forecasting the next period with the best models. Methods of data collection in this study is literature review and documentation of currency exchange agencies, in this case the bank BI, BTN and BOTM via internet. The data is taken from the daily data. Based on the analysis relative efficiency of estimators Gaussian kernel function of the polynomial estimator obtained by 0.000088. With variance and MSE of the estimator Gaussian kernel function is 0.000008886 and 0.3867, while the variance and MSE of the polynomial estimator is 0.39019 and 0.10078. It can be concluded that the estimator Gaussian kernel function is more efficient and is the best model because the variance and MSE estimator Gaussian kernel function is smaller than polynomial estimator. The best model can be used for forecasting the next. Forecasting results by using the best model for the 6th day of 82.6067.

Actions (login required)

View Item