Ahead

Univariate and multivariate time series forecasting with uncertainty quantification (including simulation).

Documentation

This package is a Julia wrapper for R package ahead, and it follows its API as closely as possible: see https://techtonique.r-universe.dev/ahead#reference for details.

Installation

julia> using Pkg; Pkg.add(path = "https://github.com/Techtonique/Ahead.jl")

Examples

using Ahead
y = [1,2,3,4,5,6,7,8,9,10]
z = rand(8, 2)
u = rand(50)

Univariate forecasting with dynrmf, Dynamic regression model

val1 = Ahead.dynrmf(y, h=6)
println(val1) 

Multivariate forecasting with ridge2f, Random Vector functional link network model with 2 regularization parameters (see https://www.mdpi.com/2227-9091/6/1/22)

val2 = Ahead.ridge2f(z, h=4)
println(val2) 

TODO

• Use TimeSeries.jl (https://juliastats.org/TimeSeries.jl/stable/)

Note to self or developers

Docker

docker build -t ahead .&&docker run ahead 

Install locally

• Install RCall: https://juliainterop.github.io/RCall.jl/stable/installation/
• Getting started with RCall: https://juliainterop.github.io/RCall.jl/latest/gettingstarted.html
• Julia RCall: https://www.geeksforgeeks.org/julia-rcall/

$ rm -rf ~/.julia/compiled # clearing cache
$ rm -rf ~/.julia/lib # clearing cache
julia> using Pkg; Pkg.rm("Ahead")
pkg> rm Ahead 
julia> using Pkg; Pkg.add(path="https://github.com/Techtonique/Ahead.jl")
julia> using Pkg; Pkg.update("Ahead")
pkg> activate Ahead