Using Turing.jl for an ARIMA 1,1,0 model.
Is there a way to extract the AR,MA (if i used 1,1,1) coefficients from the fitted model?
@model ARIMA110(x) = begin
T = length(x)
μ ~ Uniform(-10, 10)
ϕ ~ Uniform(-1, 1)
for t in 3:T
val = μ + # Drift term.
x[t-1] + # ARIMA(0,1,0) portion.
ϕ * (x[t-1] - x[t-2]) # ARIMA(1,0,0) portion.
x[t] ~ Normal(val, 1)
end
end
out_arima = ARIMA110(x)
Thanks,
Andrew