get the code actually rendered

docs/make.jl

@@ -15,6 +15,7 @@ makedocs(
     sitename = "SafetySignalDetection",
     format = Documenter.HTML(
         assets = String["assets/citations.css"],
+        size_threshold = nothing
     ),
     modules = [SafetySignalDetection],
     pages = [

docs/src/introduction.md

@@ -12,7 +12,7 @@ Note that:
 - The column `time` needs to be a `Float64` (time until adverse event or until last treatment or follow up).
 - The column `trialindex` needs to be `Int64` (index of historical trials, starting from 1 and consecutively numbered).
 
-```julia
+```@example 1
 using CSV
 using DataFrames
 using SafetySignalDetection
@@ -43,13 +43,14 @@ Note:
 - The number of samples (e.g. 10,000) and the number of threads (1) need to be passed as integers. Otherwise the call will fail.
 - We don't need to discard a burn in period now because the NUTS sampler is used internally and is discarding a burn in automatically already.
 
-```julia
+```@example 1
 using Distributions
 
 prior_a = Beta(1 / 3, 1 / 3)
 prior_b = Beta(5, 5)
 
 map_samples = meta_analytic_samples(df, prior_a, prior_b, 10_000, 1)
+map_samples[1:5]
 ```
 
 The resulting $\pi^{*}$ samples in `map_samples` are from the meta-analytic prior for the adverse event rate per unit of time in the control arm in the ongoing blinded trial.
@@ -63,7 +64,7 @@ Internally, the classic Expectation Maximization (EM) algorithm is used.
 We can check the approximation graphically by overlaying the resulting probability density function with the
 samples histogram.
 
-```julia
+```@example 1
 using Plots
 
 map_approx = fit_beta_mixture(map_samples, 2)
@@ -82,7 +83,7 @@ The important new ingredients here are:
 - The prior on the experimental adverse event rate. This can be uninformative, or weakly informative using a diluted background rate distribution.
 - The proportion of patients in the experimental treatment arm.
 
-```julia
+```@example 1
 prior_exp = Beta(1, 1)
 prior_ctrl = map_approx
 exp_proportion = 0.5
@@ -92,19 +93,20 @@ df_current = DataFrame(CSV.File(csv_current_path))
 df_current.y = map(x -> x == 1, df_current.y)
 
 result = blinded_analysis_samples(df_current, prior_exp, prior_ctrl, exp_proportion, 10_000, 1)
+first(result, 5)
 ```
 
 ## Summary
 
 We can now e.g. look at the posterior probability that the rate in the experimental arm, $\pi_E$, is larger than the event rate in the control arm, $\pi_C$.
 
-```julia
+```@example 1
 mean(result[!, :pi_exp] .> result[!, :pi_ctrl])
 ```
 
 We can also create corresponding plots.
 
-```julia
+```@example 1
 stephist(result[!, :pi_exp], label = "experimental arm", norm = :pdf)
 stephist!(result[!, :pi_ctrl], label = "control arm", norm = :pdf)
 ```