I was wondering how I can simulate data given parameter values using a turing model? Any help would be great!
ex.
using Turing, Distributions
@model function logistic_regression(x, y, n, σ)
intercept ~ Normal(0, σ)
student ~ Normal(0, σ)
balance ~ Normal(0, σ)
income ~ Normal(0, σ)
for i in 1:n
v = logistic(intercept + student * x[i, 1] + balance * x[i, 2] + income * x[i, 3])
y[i] ~ Bernoulli(v)
end
end;
I would like to generate y after setting student =1, balance = 1, intercept = 1, income = 1 and x = [1,1,1] without having to re-code the model.
I’m just using Bayesian logistic regression as an example. My actual model is hierarchical. I’m trying to do parameter recovery from simulated data to check the fitting process. Life would be much easier if I can utilize the turing model for data simulation. Functions for prior predictive checks would help if you can direct me to them but ideally I’d like functions where I can simulate data with parameters that I chose. Thanks for the suggestion though.