include ridge2f

README.md

@@ -1,8 +1,29 @@
 # Ahead
 
-[![Build Status](https://github.com/thierrymoudiki/Ahead.jl/actions/workflows/CI.yml/badge.svg?branch=main)](https://github.com/thierrymoudiki/Ahead.jl/actions/workflows/CI.yml?query=branch%3Amain)
+[![Build Status](https://github.com/Techtonique/Ahead.jl/actions/workflows/CI.yml/badge.svg?branch=main)](https://github.com/Techtonique/Ahead.jl/actions/workflows/CI.yml?query=branch%3Amain) [![HitCount](https://hits.dwyl.com/Techtonique/Aheadjl.svg?style=flat-square)](http://hits.dwyl.com/Techtonique/Aheadjl)
 
+**Univariate and multivariate time series forecasting, with uncertainty quantification**.
 
+## Installation
+
+From GitHub (for now):
+
+```julia
+julia> using Pkg; Pkg.add("https://github.com/Techtonique/Ahead.jl")
+````
+
+Ultimately:
+
+```julia
+julia> using Pkg; Pkg.add("Ahead")
+```
+
+## Usage
+
+```julia
+julia> using Ahead
+julia> print(Ahead.dynrmf([1,2,3,4,5,6,7,8,9,10]))
+```
 
 ## Note to self or developers
 

src/Ahead.jl

@@ -2,6 +2,7 @@ module Ahead
 
 	# exports 
 	export dynrmf
+	export ridge2f
 
 	if Sys.islinux()		
 		run(`ls -la`)
@@ -52,10 +53,113 @@ module Ahead
 
 	R"load_ahead <- try(library(ahead), silent = TRUE)"		
 
-	function dynrmf(x)
-		# https://juliainterop.github.io/RCall.jl/stable/custom/#Nested-conversion				
-		return rcopy(R"try(ahead::dynrmf($x), silent = TRUE)")	
+	# univariate time series forecasting models ----------------------------------------------
+
+	""" 
+	Dynamic regression model (see https://techtonique.r-universe.dev/ahead/doc/manual.html#dynrmf)
+
+	# Details
+
+	For now, the function uses only Ridge regression with automatic selection of the regularization parameter.
+
+	# Examples
+
+	```julia
+	using Ahead
+	val = Ahead.dynrmf([1,2,3,4,5,6,7,8,9,10], h = 5, level = 95)   
+	println(val)
+	````
+
+	"""
+	function dynrmf(y, h = 5, level = 95)	
+		r_code = """
+		ahead::dynrmf(y = $y, h = $h, level = $level)
+		"""					
+		return rcopy(R"$r_code")	
 	end 	
 
+	# multivariate time series forecasting models ---------------------------------------------- 
+
+	"""
+	Random Vector functional link (RVFL) nnetwork model with 2 regularization parameters (see https://techtonique.r-universe.dev/ahead/doc/manual.html#ridge2f)
+
+	# Details
+
+	The model is based on an RVFL model (see https://www.mdpi.com/2227-9091/6/1/22) with 2 regularization parameters, and provides some uncertainty quantification.
+
+	# Examples
+
+	```julia
+	using Ahead
+	using Distributions
+	import Random
+	Random.seed!(123)
+	y = rand(100, 5)
+	print(ridge2f(y, h = 5, level = 95))
+	```
+
+	""" 
+	function ridge2f(
+		y,
+		xreg = nothing,
+		h = 5,
+		level = 95,
+		lags = 1,
+		nb_hidden = 5,
+		nodes_sim = "sobol",
+		activ = "relu",
+		a = 0.01,
+		lambda_1 = 0.1,
+		lambda_2 = 0.1,
+		dropout = 0,
+		type_forecast = "recursive",
+		type_pi = "gaussian",
+		block_length = nothing,
+		margins = "gaussian",
+		seed = 1,
+		B = 100,
+		type_aggregation = "mean",
+		centers = nothing,
+		type_clustering = "kmeans",
+		ym = nothing,
+		cl = 1,
+		show_progress = true,
+		args...
+	  )
+
+	  if xreg == nothing
+		xreg = RNULL()
+	  end
+
+	  if block_length == nothing	
+		block_length = RNULL()
+	  end
+
+	  if centers == nothing
+		centers = RNULL()
+	  end
+
+	  if ym == nothing
+		ym = RNULL()
+	  end
+
+	  r_code = """
+	  ahead::ridge2f(y = $y, xreg = $xreg, h = $h, 
+	  level = $level, lags = $lags, nb_hidden = $nb_hidden, 
+	  nodes_sim = $nodes_sim, activ = $activ, a = $a, 
+	  lambda_1 = $lambda_1, lambda_2 = $lambda_2, 
+	  dropout = $dropout, type_forecast = $type_forecast,
+	  type_pi = $type_pi, block_length = $block_length, 
+	  margins = $margins, seed = $seed, B = $B, 
+	  type_aggregation = $type_aggregation, 
+	  centers = $centers, type_clustering = $type_clustering, 
+	  ym = $ym, cl = $cl, show_progress = $show_progress, 
+	  $args...)
+	  """
+
+	  return rcopy(R"$r_code")
+
+	end 
+
 end
 

test/runtests.jl

@@ -2,6 +2,19 @@ using Ahead
 using Test
 
 @testset "Ahead.jl" begin
+
+    using Ahead
+	using Distributions
+	import Random
+
+    # Test Ahead.dynrmf ----------------------------
 	val = Ahead.dynrmf([1,2,3,4,5,6,7,8,9,10])   
     @test isapprox(round(val[:residuals][1]), 0)
+
+    # Test Ahead.ridge2f ----------------------------
+	Random.seed!(123)
+	y = rand(100, 5)
+    val = Ahead.ridge2f(y, h = 5, level = 95)
+	@test val[:x] == y
+
 end