Prior predictive checks

Checking my understanding of prior predictive checks and the Wiener process model

Author

Sheng Long

Very simple toy model: y ~ N(mu, sigma).

# # data 
# df <- as_tibble(rnorm(100, 5, 3))
# # formula 
# f <- bf(value ~ 1)
# make_stancode(f, data = df)

Prior Predictive Checks

source: https://mc-stan.org/docs/stan-users-guide/posterior-predictive-checks.html#prior-predictive-checks

Let’s assume mu ~ N(4, 2) and sigma ~ N(3, 1):

# # Define the data 
# N <- 100
# my_x <- rnorm(N, 0, 1)
# 
# stan_data <- list(
#   N = N, 
#   x = my_x
# )
# 
# model <- cmdstan_model("ex1.stan")

Fit the stan model:

# fit1 <- model$sample(
#   data = stan_data, 
#   chains = 4, 
#   fixed_param = TRUE, 
#   iter_warmup = 0, 
#   iter_sampling = 1000, 
#   refresh = 0
# )

# fit1$draws("y_sim") %>% as_draws_df(.) %>% 
#   select(1) # %>% 
#   ggplot(aes(x = `y_sim[1]`)) # + 
#   geom_slab()

Prior predictive check with a Weiner process model

# TODO