Strange problems with my zero-inflated models using brms package

Interfaces brms
1

Hi Everyone,

I’m having problems to fit Zero Inflated Poisson (ZIP) and Zero Inflated Negative Binomial (ZIBN) models using the brms package.

For example:

ZIP.noCAR.prior = c((prior(normal(0,1000),class = b)),
                    (prior(normal(0,1000),class = Intercept)))

ZIP.ind <- brm(brm(formula = Fevereiro ~ PIB_1000 +
                     offset(log(Pop)),zi ~ TOTAL),
               data = saopaulodf,
               prior = ZIP.noCAR.prior,
               family = zero_inflated_poisson(),
               iter = 50000,
               warmup = 10000,
               thin = 50,
               chains = 1,
               init = 0,
               control = list(max_treedepth=12))

And this is the error message I got :

Error: Argument 'data' must be coercible to a data.frame.

But,


> class(saopaulodf)
[1] "data.frame"

And when I use a CAR component like this code below:

ZIP.prior = c((prior(normal(0,1000),class = b)),
              (prior(normal(0,1000),class = Intercept)),
              (prior(uniform(0, 1000),class= sdcar)),
              (prior(beta(1, 1),class= rhocar))
)
ZIP <- brm(bf(brm(formula = Fevereiro ~ PIB_1000 +
                    offset(log(Pop)) + car(W,type = "bym2")),zi ~ TOTAL),
           data = saopaulodf,
           data2 = list(W = W),
           prior = ZIP.prior,
           family = zero_inflated_poisson(),
           iter = 50000,
           warmup = 10000,
           thin = 50,
           chains = 1,
           init = 0,
           control = list(max_treedepth=12))

I got this error message:

Error: Object 'W' was not found in 'data2'.

But when I make a Poisson model using the same objects I can fit the model with no problems. For example:


poi.prior = c((prior(normal(0,1000),class = b)),
              (prior(normal(0,1000),class = Intercept)),
              (prior(gamma(1, 1),class= sdcar)),
              (prior(beta(1, 1),class= rhocar))
)
poi <- brm(formula = Fevereiro ~ PIB_1000 +
                               car(W,type = "bym2"),
                         data = saopaulodf,
                         data2 = list(W = W),
                         prior = poi.prior,
                         family = poisson(),
                         iter = 50000,
                         warmup = 10000,
                         thin = 50,
                         chains = 1,
                         init = 0,
                         control = list(max_treedepth=12))

Can anyone help me with this issue ?

  • Operating System: Windows 11; R 4.4.1
  • brms Version:2.22.0
2

You have typo there. it should be brm(bf(

2 Likes