PEMILIHAN MODEL UNTUK REGRESI KUANTIL TOBIT DENGAN MENGGUNAKAN GIBBS SAMPLING; MODEL SELECTION IN TOBIT QUANTILE REGRESSION USING THE GIBBS SAMPLING

Bintan, Fadhilah Fiqih Kus Alfa, Sri Haryatmi

2016 | Skripsi | FMIPA

Tobit quantile regression can be used to overcome the limitation of regression to analyze censored data which not symmetric and outlier existed. Tobit quantile regression can be estimated by using bayesian method which based on the information derived from sample and prior information. The combination of sample and prior information is called by posterior. So difficult to find a posterior distribution wich include by many parameters. Therefore, there is a special techniques which will make easily, that can be uses by Gibbs Sampling. R software provides a Brq package for doing analysis in tobit quantile regression using Gibbs Sampling completely. The case study in this paper discussesthe factors that effect people become farmer. The estimation result of tobit quantile regression will be compared with logistic regression, and compared with tobit regression method. By using value of MSE provide a conclusion that estimation of tobit quantile regression more accurate and precision than other estimation.

Kata Kunci : Tobit Quantile Regression, Bayesian, Gibbs sampling, Asymmetric Laplace distribution