Markov Zinciri Monte Carlo Yönteminin Dinamik Doğrusal Modellere Uygulanması

Abstract

This study deals with implementing three different types of dynamic linear models using dlm package in R. These models are the steady model, the linear growth model and the combined model as a second order polynomial model with a seasonal effect. Several methods such as maximum likelihood, forward filtering backward sampling and or Gibbs sampling are applied for each model considered in the study. Datasets are generated for the first two models; Turkey Cost of Living Index (Wage Earners) series is used for the combined model. The estimations of the unknown variance components besides, the Kalman filter results are obtained in the study.