|
|
 |
| |
|
|
|
Multilevel Bayesian Regression Model with some Selected Horseshoe Flat Priors for Count and Continuous Responses |
|
|
|
PP: 137-150 |
|
|
|
doi:10.18576/jsapl/100302
|
|
|
|
Author(s) |
|
|
|
Rasaki Olawale Olanrewaju,
Sodiq Adejare Olanrewaju,
|
|
|
|
Abstract |
|
|
| This article provides both the theoretical and experimental analysis for Bayesian multilevel regression model via fully specified random and fixed effects, coupled with variance error. The dependent variables for the multilevel Bayesian regression that are usually supported by distributions that are regarded as horseshoe priors for Bayesian multilevel regression analysis. The classes of horseshoe priors for multilevel regression used in this research are referred to as horseshoe flat priors for count and continuous regression multilevel responses that might be via non-linear and linear multilevel regression models with options for covariance structures, autocorrelation of covariates or responses. Among the horseshoe flat prior used to estimate the corresponding posteriors regression means and predictive checks are Binomial, Poisson, and Negative-Binomial; and Gamma, Lognormal for count and continuous responses respectively. |
|
|
|
|
|
|
|
