Nonlinear model with grouped data and a single control, model specification problem in BRMS

Modeling
, ,
2

Oh, maybe I’ve found something wrong. It is to do with the model specification:

prior1 <- prior(normal(14, 1), nlpar = "top") + 
  prior(normal(8.0, 1), nlpar = "bottom")+         
  prior(normal(0.2, 0.1), nlpar = "rate")+
  prior(normal(2, 1), nlpar = "KG5"  )

# prior(normal(0.5, 1), nlpar = "Group"  )+

fit1 <- brm(bf(Response ~ top - ((top-bottom)*exp(-rate * Treatment)) + KG5, 
               top + bottom + rate ~ 1, 
               KG5 ~ 1 + Control, # < change here, now can add factor 
               nl = TRUE),        # drawing from "control" column
            data = SynthData, 
            prior = prior1,
            #family = cumulative(link="logit", threshold="flexible"),
            iter = 4000 , warmup = 2000, thin = 2,
            chains = 4, cores=6,
            init = init_list, 
            control = control,
            #file = "./model/fit14"
)

Visualize

testPred <- predict(
  fit1,
  type = "response")

SynthData$Estimate <- testPred[,1]

#Visualize
ggplot(SynthData, aes(x=Treatment, y= Response))+
  geom_point(aes(color = Group, shape = Control))+
  geom_point(aes(x=Treatment, y=Estimate, color = Group, shape = Control), size = 6)+
  theme_bw()

image
image748×843 7.19 KB

So i can now see the control group correctly fitting.
My knowledge on formulating these BRMS models is a bit hit and miss. Does anyone know of a good resource that tackles this in a systematic way?

Thomas