Help with misspecified model

Modeling
,
1

data {
int<lower=0> N;//Number of observations
int<lower=1> J;//Number of predictors with random slope
int<lower=1> K;//Number of predictors which are fixed effects
int<lower=1> L;//Number of customers/groups
int<lower=0,upper=1> y[N];//Binary response variable
int<lower=1,upper=L> ll[N];//Number of observations in groups
matrix[N,K] x1;
matrix[N,J] x2;
}
transformed data {
vector[J] ones = rep_vector(1, J);
}
parameters {
row_vector[J] rbeta_mu; //mean of distribution of beta parameters
row_vector<lower=0>[J] rbeta_sigma; //variance of distribution of beta parameters
row_vector[J] beta_raw[L]; //group-specific parameters beta
vector[K] beta;
}
transformed parameters {
matrix[L,J] rbeta;
for (l in 1:L)
rbeta[l] = rbeta_mu + rbeta_sigma .* beta_raw[l]; // coefficients on x
}
model {
vector[N] p;
rbeta_mu ~ normal(0,5);
rbeta_sigma ~ inv_gamma(1,1);
beta~normal(0,5);
for (l in 1:L)
beta_raw[l] ~ std_normal();

p = x1 * beta + (x2 .* rbeta[ll]) * ones; // Multiplication by vector of ones as a row-wise summation of matrix
y~bernoulli_logit§;
}

I am trying to fit a fairly simple multitlevel model. However I feel the model might be misspecified since when i tried to run the model using cmdstanR i get lot of errors.

image
image1380×928 423 KB

I am a newbie to Bayesian statistics. So I am looking for help from the community to determine what is the mistake.

Thanks in advance!

2

Welcome to the Stan forum!

the warning says that the value of the probability parameter is nan, and the relevant section of your model reads

y~bernoulli_logit§;

this should be changed to

y~bernoulli_logit(p);

In case your model already correctly uses y~bernoulli_logit(p);: The warning message is not a concern if it comes up at the beginning. (i.e. nor further warnings later during warm-up and especially during sampling)

Also, your priors look fairly wide (assuming the values in x1 and x2 are not very small).
I would recommend to

  • start with narrower priors (e.g. normal(0,2) throughout, even for rbeta_sigma)
  • do a prior predictive check, i.e. comment out // y~bernoulli_logit(p);, run the model, and look at the distribution of the parameter p values.

Lastly, if you put
"``` Stan
[your stan model]
"````

around your Stan model, it will be easier to read for others

3

Hi @Guido_Biele, It actually does use it as y~bernoulli_logit( p ). I don’t why it appeared here as so.

I will try prior predictive check and narrower priors. Thank you !