Automatic multi-seasonal ARIMA Learning with Box and Jenkins method.
using AutoARIMA
julia> z = seriesG
144-element StaticArrays.SVector{144, Int64} with indices SOneTo(144):
112
118
132
129
⋮
508
461
390
432
julia> sarima = fit(MSARIMAParams([0,0],[1,1],[1,1],[1,12]),log.(z))
ARMAModel{13, 13, Float64}(-0.002232109177566169, [1.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, 1.0, -1.0], [0.34919600156185515, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, 0.31610711882734377, -0.11038334195974667], Float64[], 0.0016285014276129083)
julia> forecast(sarima,log.(z))
6.104360295754863
julia> best = boxjenkins(z,true,2,2,2,[1,12],criterium=aicc)
AutoARIMA.ARMAXModel{25, 0, 0, Float64}(-0.0009527614685900121, [1.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0 … -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -0.0, -1.0, 1.0], Float64[], Float64[], 0.004332904404300882)