Weakly informative priors

This model is a continuation of the Wild Chain.

Data

y = [-1,1]

Model

using Turing

@model function m8_3(y)
    α ~ Normal(1, 10)
    σ ~ truncated(Cauchy(0, 1), 0, Inf)
    μ = α

    y ~ Normal(μ, σ)
end;

Output

chns = sample(m8_3(y), NUTS(), 1000)

using StatsPlots

StatsPlots.plot(chns)

Original output

"""
      mean   sd  5.5% 94.5% n_eff Rhat
alpha 0.09 1.63 -2.13  2.39   959    1
sigma 2.04 2.05  0.68  4.83  1090    1
""";