Skip to content
Prev 13920 / 20628 Next
Original Raw

Linear mixed model - heterogeneity

Paul Buerkner 
You may also try the brms package, which has a hurdle_gamma family that
might be helpful to you.

A sample hurdle_gamma model (using the epilepsy data of brms) may look like
this:

fit <- brm(count ~ 0 + trait * (log_Age_c + log_Base4_c * Trt_c)
            + (0+trait||patient),
            data = epilepsy, family = hurdle_gamma("log"))

The reserved variable "trait" has to levels, one for the gamma part and one
for the bernoulli part modeling zeros.

Currently, hurdle_gamma models are only available in the github version of
brms to be installed via

library(devtools)
install_github("paul-buerkner/brms")

Since brms is based on Stan, you will need a C++ compiler. Instructions on
how to get one are presented at the end of the README on
https://github.com/paul-buerkner/brms.



2015-10-23 16:15 GMT+02:00 Etn bot <etnbot1 at gmail.com>:

  
  


  


  


  
  

    

  

    
      
    
    
Thread (3 messages)

  

  

    
            
        Etn bot
        
        Linear mixed model - heterogeneity
      
      Oct 23
    
    
      
            
        Ben Bolker
        
        Linear mixed model - heterogeneity
      
      Oct 29
    
    
      
            
        Paul Buerkner
        
        Linear mixed model - heterogeneity
      
      Oct 30