Alejandro De La Fuente - 2024-01-25

Hi,

I need some help calculating the 90th percentile of a 3D vector within a JAGS model. This needs to be done to integrate 2 models, so that the output of the first model is used as a predictor in the second model. I am providing an example below of how could it be done using the mean, but I am interested in calculating the 90th percetile across one of the vector's dimension instead. Has anyone run into a similar situation before or know how to do this? I have been searching through the manual, but I could not figure out a way to achieve this.

Thank you very much in advance for your time.

All my best,

Alejandro

Example:

Model 1: the response variable is contained within a 3D array

for( i in 1:I ) {
    for( j in 1:J ) {
        for( z in 1:Z ) {

            y[i,j,z] ~ dnorm(mu, tau)

            }
         }
      }

Derived quantities from model 1. Summarising y across the z dimension. Here is my issue, I need to calculate the 90th percentile instead of the mean.

I then need to use the variable y[i,j,z] as a predictor in a model of only 1 dimension. To do that, I need to summarise the y variable across the dimension of interest. I am showing next an example taking the average.

for( z in 1:Z){

    y.mean[z] <- mean(y[,,z])

 }

Model 2: using derived quantities from model 1 as a predictor in model 2

Then, I will use y.mean as a predictor in the next model. All this is done within one single hierarchical model, so the uncertainty is transferred.

for( z in 1:Z ){

    y2[z] ~ dnorm(mu2[z], tau2)
    mu2[z] <- a + b * y.mean[z]

 }

END OF EXAMPLE