PAR.jl

Repository with Periodic Autorregressive models.

The main goal of the repository is to implement Periodic Autorregressive (PAR) models and some simple variations of the model. The full methodology of PAR models can be found in this article (Maceira, Maria & Penna, Débora & Damazio, Jorge. (2006). Geração de Cenários Sintéticos de Energia e Vazão para o Planejamento da Operação Energética. Cadernos do IME : Série Estatística. 21. 10.12957/cadest.2006.15760.) a link to the article's pdf can be found here (Available only in Portuguese.)

To add the package you can do:

add https://github.com/psrenergy/PAR.jl.git

and to run an example you can do:

using PAR
funil_grande = include(joinpath(pkgdir(PAR), "test", "data", "funil_grande.jl"))
batalha = include(joinpath(pkgdir(PAR), "test", "data", "batalha.jl"))

n_stages = 12
p_lim = 6
par_1 = PARp(funil_grande, n_stages, p_lim; information_criteria = "aic");
par_2 = PARp(batalha, n_stages, p_lim; information_criteria = "aicc");
fit_par!(par_1);
fit_par!(par_2);

steps_ahead = 100
n_scenarios = 1000
scen = simulate_par([par_1; par_2], steps_ahead, n_scenarios)