ANALISIS BAYESIAN UNTUK REGRESI SPLINE TERPENALTI DENGAN ERROR YANG HETEROSKEDASTIS; (BAYESIAN ANALYSIS FOR PENALIZED SPLINE REGRESSION WITH HETEROSCEDASTIC ERRORS)

Fitriani, Rika, Gunardi

2015 | Disertasi | FMIPA

In this thesis, Bayesian analysis for penalized spline regression with heteroscedastic errors will be discussed. Penalized spline regression with heteroscedastic errors parameters can be estimated by Gibbs sampling and Metropolis Hasting approach. Bayesian analysis for penalized spline regression with heteroscedastic errors will be applied to analyze the relationship between the Rupiah exchange rate and the Jakarta Composite Index (JCI). The estimation result of Bayesian analysis for penalized spline regression with heteroscedastic errors is compared to Bayesian analysis for penalized spline regression with homoscedastic errors and simple linear regression. The estimation result of Bayesian analysis for penalized spline regression with heteroscedastic errors has the smallest MSE and the largest R2.

Kata Kunci : penalized spline regression; heteroscedastic errors; Gibbs sampling; Metropolis Hasting