Redefining a vector for use in a hierarchical model

Modeling
1

Hi all,

I’m fitting a piecewise regression model with a random transition point, tp.
I’m doing this based on data, x_data, which is defined as a vector of length N.

I’ve successfully used to below code to define a variable that tells me whether my data is before this transition point (region_1) or after it, (region_2):

parameters{
  
...
  
  real tp; // transition point  

...

}

transformed parameters {
  
// Introduce an indicator variable to distinguish between the 2 sides of the transition point
  
  vector [N] region_2; 

  for (i in 1:N){
    
    if (x_data[i] < tp) {
        region_2[i] = 0;
    } else {
        region_2[i] = 1;
    }
  }
}

I wanted to develop this into a hierarchical model, with different model parameters (including tp) for each different group. I need a version of my region_2 vector to work for this case.

My attempts to do this have so far not resulted in any samples. I believe there is a problem with how I am indexing my new region_2 parameter.

Please let me know if you have any experience or advice. Thanks!

2

Hi all,

The below code indexes the transition point, tp and region_2 parameters in the way I was hoping for:

transformed parameters {
  
// Introduce an indicator variable to distinguish between the 2 sides of the transition point
  
  vector [N] region_2;

  for (i in 1:N){
    
    if (x_data[i] < tp[test[i]]) {
      region_2[i] = 0;
    } else {
      region_2[i] = 1;
    }
  }
}

Apologies if I did not explain myself clearly in the above question.

1 Like