Good evening.
I was wondering if there is any paper or article that explains how to obtain a Poisson-Gamma Mixture model for count data using the method MCMC with the M-H algorithm in Stata. More precisely, if there is an article that includes spatial random effects using a SWM. Now I'm trying to perform that kind of regression with disastrous results (mainly for the code).
Just want to know if anyone has done it before starting a full discussion. My code:
bayesmh y= (exp({eta: x Phi[_ID], noconstant})), ///
mcmcsize(12500) burnin(2500) initrandom saving(example, replace) ///
likelihood(poisson) ///
prior({xb_y}, gamma({k},{k}/{eta:})) ///
block({xb_y}, split) ///
prior({rho}, uniform(-1,1)) ///
block({rho}, split) ///
prior({tao}, gamma(0.001,000.1)) ///
block({tao}, split) ///
prior({eta: x}, uniform(-0.1,0.1)) ///
block({eta: x}, split) ///
prior({eta: Phi[_ID]}, normal(0,1)) ///
block({eta: Phi[_ID]}, split) ///
prior({k}, gamma(4,1)) ///
block({k}, split)
Thanks for your time.
Best regards.
I was wondering if there is any paper or article that explains how to obtain a Poisson-Gamma Mixture model for count data using the method MCMC with the M-H algorithm in Stata. More precisely, if there is an article that includes spatial random effects using a SWM. Now I'm trying to perform that kind of regression with disastrous results (mainly for the code).
Just want to know if anyone has done it before starting a full discussion. My code:
bayesmh y= (exp({eta: x Phi[_ID], noconstant})), ///
mcmcsize(12500) burnin(2500) initrandom saving(example, replace) ///
likelihood(poisson) ///
prior({xb_y}, gamma({k},{k}/{eta:})) ///
block({xb_y}, split) ///
prior({rho}, uniform(-1,1)) ///
block({rho}, split) ///
prior({tao}, gamma(0.001,000.1)) ///
block({tao}, split) ///
prior({eta: x}, uniform(-0.1,0.1)) ///
block({eta: x}, split) ///
prior({eta: Phi[_ID]}, normal(0,1)) ///
block({eta: Phi[_ID]}, split) ///
prior({k}, gamma(4,1)) ///
block({k}, split)
Thanks for your time.
Best regards.
