mcmc - Jags/Bugs one step ahead prediction -

Imagine a simple development model How do I get a forward-forward forecast?

  # East and obstacles N.est [1] ~ Doñif (0, 10) # The size of the initial population means mean.lambda ~ dunif (0, 10) #internal.ps i ~ Doinif (0, 10) #sigma Proc.-Donieff (0, 10) # sigma2.proc for state of the process process & lt; - Po (sigma prok, 2) tau.proc & lt; - Pa (sigma.proc, -2) sigma.obs ~ dunif (0, 10) # observation process sigma2.obs & lt; - pow (sigma.obs, 2) tau.obs & lt; - POW before (sigma.ob, -2) sigma.pcI ~ doinif (0, 10) #section procedure sdi for ssi 2 sps & lt; - Po (Sigma SC, 2) Tau.PC & lt; - POS (SigmaAPS), -2) State Process for Chance # 1 (1: (T1) in T {lambda [t] ~ dornum (meaning: lambda, tau prok) psi [t] ~ denom (Mean.pc, Tau Psi) NEST [T + 1] & lt; - 10 / (1 + XP (- PSI [T] * (nest [t] * lambda [t])) #Neast [t] * lambda [t] + sai [t]} # for the observation process ( 1: T in T) {y [t] ~ dnorm (NESST, T.O.O.A.)}  

From a book, this example. How do I get a step ahead of the forecast, because I am not interested in the smooth output.

Do you want to set up units that you want to change the response vector y Want to predict from to NA and then want to create a duplicate sample of y . ;

Checking residues between y.pred [t] ~ dnorm (n.st [t], towbs)

y For and y.pred

res < - y - y.pred