Specifying prior parameter as input in data section?

Questions
1

Is there any difference between specifying priors manually in code (model 1) or as input in the data section (model 2)?
I am asking because model 1 allows me to change the prior when running the stan code while model 2 means I have to change the priors in the stan code and save them manually. Any idea if there is a difference and if there is, what is going on?

Model 1:
data {
int <lower=0> AG;
int <lower=0> rep_mean;
}
model {
reporting_rate0 ~ normal(rep_mean, 1);
}

Model 2:
data {
int <lower=0> AG;
int <lower=0> rep_mean;
}
model {
reporting_rate0 ~ normal(-2, 1);
}

2

Hi @sk8. I think you already answered your question?
It’s not clear if your question is really about Stan or not, what goes on behind the scenes might differ between Stan and PyMC.

3

Hi @drbenvincent, indeed my questions is about Stan, I’ll post it on the right forum. Thanks!

1 Like