wip

Author: BatyLeo

Project.toml

@@ -25,6 +25,7 @@ SCIP = "82193955-e24f-5292-bf16-6f2c5261a85f"
 SimpleWeightedGraphs = "47aef6b3-ad0c-573a-a1e2-d07658019622"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
+StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 
 [compat]
 ConstrainedShortestPaths = "0.6.0"
@@ -47,6 +48,7 @@ Requires = "1.3.0"
 SimpleWeightedGraphs = "1.4"
 SparseArrays = "1"
 Statistics = "1.11.1"
+StatsBase = "0.34.4"
 julia = "1.6"
 
 [extras]

src/PortfolioOptimization/PortfolioOptimization.jl

@@ -5,7 +5,7 @@ using DocStringExtensions: TYPEDEF, TYPEDFIELDS, TYPEDSIGNATURES
 using Distributions: Uniform, Bernoulli
 using Flux: Chain, Dense
 using Ipopt: Ipopt
-using JuMP: @variable, @objective, @constraint, optimize!, value
+using JuMP: @variable, @objective, @constraint, optimize!, value, Model
 using LinearAlgebra: I
 using Random: MersenneTwister
 

src/Utils/Utils.jl

@@ -7,6 +7,7 @@ using JuMP: Model
 using LinearAlgebra: dot
 using SCIP: SCIP
 using SimpleWeightedGraphs: SimpleWeightedDiGraph
+using StatsBase: StatsBase
 
 include("data_sample.jl")
 include("interface.jl")
@@ -24,5 +25,6 @@ export grid_graph, get_path, path_to_matrix
 export neg_tensor, squeeze_last_dims, average_tensor
 export scip_model, highs_model
 export objective_value
+export compute_normalizer
 
 end

src/Utils/data_sample.jl

@@ -18,3 +18,44 @@ $TYPEDFIELDS
     "instance object (optional)"
     instance::I = nothing
 end
+
+function _transform(t, sample::DataSample; kwargs...)
+    (; instance, x, θ_true, y_true) = sample
+    return DataSample(; instance, x=StatsBase.transform(t, x; kwargs...), θ_true, y_true)
+end
+
+function _reconstruct(t, sample::DataSample; kwargs...)
+    (; instance, x, θ_true, y_true) = sample
+    return DataSample(; instance, x=StatsBase.reconstruct(t, x; kwargs...), θ_true, y_true)
+end
+
+"""
+$TYPEDSIGNATURES
+
+Compute the mean and standard deviation of the features in the dataset.
+"""
+function StatsBase.fit(transform_type, dataset::AbstractVector{<:DataSample}; kwargs...)
+    x = hcat([d.x for d in dataset]...)
+    return StatsBase.fit(transform_type, x; kwargs...)
+end
+
+"""
+$TYPEDSIGNATURES
+
+Transform the features in the dataset.
+"""
+function StatsBase.transform(t, dataset::AbstractVector{<:DataSample})
+    return map(dataset) do d
+        (; instance, x, θ_true, y_true) = d
+        DataSample(; instance, x=StatsBase.transform(t, x), θ_true, y_true)
+    end
+end
+
+# TODO: reconstruct, transform!, reconstruct!
+
+function StatsBase.reconstruct(t, dataset::AbstractVector{<:DataSample})
+    return map(dataset) do d
+        (; instance, x, θ_true, y_true) = d
+        DataSample(; instance, x=StatsBase.reconstruct(t, x), θ_true, y_true)
+    end
+end