Selection More Important Variables on D. Dimer by Using Bayesian Lasso Quantile Regression


The study aims to Prior distributions play an important role in the estimations of Bayesian theory, It is responsible for the type of Posterior Distributions for the parameters to be estimated. It is known that when the process of selecting important variables is carried out in the regression models using the Bayesian Lasso (least absolute shrinkage and selection operator) technique, the Laplace distribution must be used as a prior distribution. But using the Laplace distribution directly makes the Markov Chain Monte Carlo (MCMC) algorithm very difficult and inefficient, and it requires a very large time to converge between the estimated parameters during the iterations of estimating those parameters. In order to overcome this problem in the current paper, two transformation for Laplace distribution has been employed, which ensures that the MCMC algorithm is efficient and quick to converge between the estimated parameters within fewer iterations. Quantile Regression has good features that make it one of the distinguished models in representing the effect relationship between the dependent variable and a set of independent variables. In the current paper, the relationship between the D. dimer as a dependent variable and a set of independent variables will be modeled using the quantile regression model and estimating the parameters of this model using the Bayesian (Lasso) technique.