Return raw data instead of CounterfactualData

src/synthetic/blobs.jl → src/Raw/synthetic/blobs.jl

@@ -3,14 +3,17 @@
 
 Loads overlapping synthetic data.
 """
-function load_blobs(n=250; seed=data_seed, k=2, centers=2, kwrgs...)
+function load_blobs_raw(n=250; seed=data_seed, k=2, centers=2, kwrgs...)
     if isa(seed, Random.AbstractRNG)
         X, y = MLJBase.make_blobs(n, k; centers=centers, rng=seed, kwrgs...)
     else
         Random.seed!(seed)
         X, y = MLJBase.make_blobs(n, k; centers=centers, kwrgs...)
     end
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
-    counterfactual_data.X = Float32.(counterfactual_data.X)
-    return counterfactual_data
+
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
+    # return counterfactual_data
+
+    return (X, y)
 end

src/synthetic/circles.jl → src/Raw/synthetic/circles.jl

@@ -3,14 +3,17 @@
 
 Loads synthetic circles data.
 """
-function load_circles(n=250; seed=data_seed, noise=0.15, factor=0.01)
+function load_circles_raw(n=250; seed=data_seed, noise=0.15, factor=0.01)
     if isa(seed, Random.AbstractRNG)
         X, y = MLJBase.make_circles(n; rng=seed, noise=noise, factor=factor)
     else
         Random.seed!(seed)
         X, y = MLJBase.make_circles(n; noise=noise, factor=factor)
     end
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
-    counterfactual_data.X = Float32.(counterfactual_data.X)
-    return counterfactual_data
+
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
+    # return counterfactual_data
+
+    return (X, y)
 end

src/Raw/synthetic/linearly_separable.jl

@@ -0,0 +1,13 @@
+"""
+    load_linearly_separable(n=250; seed=data_seed)
+
+Loads linearly separable synthetic data.
+"""
+function load_linearly_separable_raw(n=250; seed=data_seed)
+    # counterfactual_data = load_blobs(n; seed=seed, centers=2, cluster_std=0.5)
+    # return counterfactual_data
+
+    raw_data = load_blobs(n; seed=seed, centers=2, cluster_std=0.5)
+
+    return raw_data
+end

src/synthetic/moons.jl → src/Raw/synthetic/moons.jl

@@ -3,14 +3,17 @@
 
 Loads synthetic moons data.
 """
-function load_moons(n=250; seed=data_seed, kwrgs...)
+function load_moons_raw(n=250; seed=data_seed, kwrgs...)
     if isa(seed, Random.AbstractRNG)
         X, y = MLJBase.make_moons(n; rng=seed, kwrgs...)
     else
         Random.seed!(seed)
         X, y = MLJBase.make_moons(n; kwrgs...)
     end
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
-    counterfactual_data.X = Float32.(counterfactual_data.X)
-    return counterfactual_data
+
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
+    # return counterfactual_data
+
+    return (X, y)
 end

src/Raw/synthetic/multi_class.jl

@@ -0,0 +1,14 @@
+"""
+    load_multi_class(n=250; seed=data_seed)
+
+Loads multi-class synthetic data.
+"""
+function load_multi_class_raw(n=250; seed=data_seed, centers=4)
+    # counterfactual_data = load_blobs(n; seed=seed, centers=centers, cluster_std=0.5)
+
+    # return counterfactual_data
+
+    raw_data = load_blobs_raw(n; seed=seed, centers=centers, cluster_std=0.5)
+
+    return raw_data
+end

src/Raw/synthetic/overlapping.jl

@@ -0,0 +1,14 @@
+"""
+    load_overlapping(n=250; seed=data_seed)
+
+Loads overlapping synthetic data.
+"""
+function load_overlapping_raw(n=250; seed=data_seed)
+    # counterfactual_data = load_blobs(n; seed=seed, centers=2, cluster_std=2.0)
+
+    # return counterfactual_data
+
+    raw_data = load_blobs_raw(n; seed=seed, centers=2, cluster_std=2.0)
+
+    return raw_data
+end

src/tabular/adult.jl → src/Raw/tabular/adult.jl

@@ -12,7 +12,7 @@ Load and preprocesses data from the UCI 'Adult' dataset
 # Example
 data = load_uci_adult(20) # loads and preprocesses 20 samples from the Adult dataset
 """
-function load_uci_adult(n::Union{Nothing,Int}=1000)
+function load_uci_adult_raw(n::Union{Nothing,Int}=1000)
     # Throw an exception if n < 1:
     if !isnothing(n) && n < 1
         throw(ArgumentError("n must be >= 1"))
@@ -45,22 +45,22 @@ function load_uci_adult(n::Union{Nothing,Int}=1000)
     )
 
     # Preprocessing
-    transformer = Standardizer(; count=true)
+    transformer = MLJModels.Standardizer(; count=true)
     mach = MLJBase.fit!(machine(transformer, df[:, DataFrames.Not(:target)]))
     X = MLJBase.transform(mach, df[:, DataFrames.Not(:target)])
     X = Matrix(X)
     X = permutedims(X)
     X = Float32.(X)
 
     y = df.target
-    counterfactual_data = CounterfactualData(X, y)
+    # counterfactual_data = CounterfactualData(X, y)
 
-    # Undersample:
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
+    # # Undersample:
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
 
-    return counterfactual_data
+    return (X, y)
 end

src/tabular/california_housing.jl → src/Raw/tabular/california_housing.jl

@@ -3,7 +3,7 @@
 
 Loads and pre-processes California Housing data.
 """
-function load_california_housing(n::Union{Nothing,Int}=5000)
+function load_california_housing_raw(n::Union{Nothing,Int}=5000)
 
     # check that n is > 0
     if !isnothing(n) && n <= 0
@@ -21,14 +21,15 @@ function load_california_housing(n::Union{Nothing,Int}=5000)
 
     # Counterfactual data:
     y = Int.(df.target)
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
-    counterfactual_data.X = Float32.(counterfactual_data.X)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
 
     # Undersample:
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
-    return counterfactual_data
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
+
+    return (X, y)
 end

src/tabular/credit_default.jl → src/Raw/tabular/credit_default.jl

@@ -3,7 +3,7 @@
 
 Loads and pre-processes UCI Credit Default data.
 """
-function load_credit_default(n::Union{Nothing,Int}=5000)
+function load_credit_default_raw(n::Union{Nothing,Int}=5000)
 
     # Load:
     df = CSV.read(joinpath(data_dir, "credit_default.csv"), DataFrames.DataFrame)
@@ -24,17 +24,17 @@ function load_credit_default(n::Union{Nothing,Int}=5000)
 
     # Counterfactual data:
     y = df.target
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(
-        X, y; features_categorical=features_categorical
-    )
-    counterfactual_data.X = Float32.(counterfactual_data.X)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(
+    #     X, y; features_categorical=features_categorical
+    # )
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
 
     # Undersample:
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
 
-    return counterfactual_data
+    return (X, y)
 end

src/tabular/german_credit.jl → src/Raw/tabular/german_credit.jl

@@ -13,7 +13,7 @@ Loads and pre-processes UCI German Credit data.
 data = load_german_credit(500) # loads and preprocesses 500 samples from the German Credit dataset
 
 """
-function load_german_credit(n::Union{Nothing,Int}=nothing)
+function load_german_credit_raw(n::Union{Nothing,Int}=nothing)
     # Throw an exception if n > 1000:
     if !isnothing(n) && n > 1000
         throw(ArgumentError("n must be <= 1000"))
@@ -37,14 +37,14 @@ function load_german_credit(n::Union{Nothing,Int}=nothing)
 
     # Counterfactual data:
     y = df.target
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
 
     # Undersample:
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
 
-    return counterfactual_data
+    return (X, y)
 end

src/tabular/gmsc.jl → src/Raw/tabular/gmsc.jl

@@ -3,7 +3,7 @@
 
 Loads and pre-processes Give Me Some Credit (GMSC) data.
 """
-function load_gmsc(n::Union{Nothing,Int}=5000)
+function load_gmsc_raw(n::Union{Nothing,Int}=5000)
 
     # Load:
     df = CSV.read(joinpath(data_dir, "gmsc.csv"), DataFrames.DataFrame)
@@ -17,15 +17,15 @@ function load_gmsc(n::Union{Nothing,Int}=5000)
 
     # Counterfactual data:
     y = df.target
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
-    counterfactual_data.X = Float32.(counterfactual_data.X)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(X, y)
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
 
     # Undersample:
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
 
-    return counterfactual_data
+    return (X, y)
 end

src/vision/cifar_10.jl → src/Raw/vision/cifar_10.jl

@@ -13,20 +13,20 @@ Loads and preprocesses data from the CIFAR-10 dataset for use in counterfactual
 data = load_cifar_10(1000) # loads and preprocesses 1000 samples from the CIFAR-10 dataset
 
 """
-function load_cifar_10(n::Union{Nothing,Int}=nothing)
+function load_cifar_10_raw(n::Union{Nothing,Int}=nothing)
     X, y = MLDatasets.CIFAR10()[:] # [:] gives us X, y
     X = Flux.flatten(X)
     X = X .* 2 .- 1 # normalization between [-1, 1]
     y = MLJBase.categorical(y)
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(
-        X, y; domain=(-1.0, 1.0), standardize=false
-    )
-    if !isnothing(n)
-        counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
-            counterfactual_data, n
-        )
-    end
-    return counterfactual_data
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(
+    #     X, y; domain=(-1.0, 1.0), standardize=false
+    # )
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
+    return (X, y)
 end
 
 """
@@ -41,13 +41,13 @@ Loads and preprocesses test data from the CIFAR-10 dataset for use in counterfac
 test_data = load_cifar_10_test() # loads and preprocesses test data from the CIFAR-10 dataset
 
 """
-function load_cifar_10_test()
+function load_cifar_10_test_raw()
     X, y = MLDatasets.CIFAR10(:test)[:]
     X = Flux.flatten(X)
     X = X .* 2 .- 1 # normalization between [-1, 1]
     y = MLJBase.categorical(y)
-    counterfactual_data = CounterfactualExplanations.CounterfactualData(
-        X, y; domain=(-1.0, 1.0)
-    )
-    return counterfactual_data
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(
+    #     X, y; domain=(-1.0, 1.0)
+    # )
+    return (X, y)
 end

src/Raw/vision/fashion_mnist.jl

@@ -0,0 +1,39 @@
+"""
+    load_fashion_mnist(n::Union{Nothing,Int}=nothing)
+
+Loads and prepares FashionMNIST data.
+"""
+function load_fashion_mnist_raw(n::Union{Nothing,Int}=nothing)
+    X, y = MLDatasets.FashionMNIST(:train)[:]
+    X = Flux.flatten(X)
+    X = X .* 2.0f0 .- 1.0f0
+    y = MLJBase.categorical(y)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(
+    #     X, y; domain=(-1.0, 1.0), standardize=false
+    # )
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
+    # # Undersample:
+    # if !isnothing(n)
+    #     counterfactual_data = CounterfactualExplanations.DataPreprocessing.subsample(
+    #         counterfactual_data, n
+    #     )
+    # end
+    return (X, y)
+end
+
+"""
+    load_fashion_mnist_test()
+
+Loads and prepares FashionMNIST test data.
+"""
+function load_fashion_mnist_test_raw()
+    X, y = MLDatasets.FashionMNIST(:test)[:]
+    X = Flux.flatten(X)
+    X = X .* 2.0f0 .- 1.0f0
+    y = MLJBase.categorical(y)
+    # counterfactual_data = CounterfactualExplanations.CounterfactualData(
+    #     X, y; domain=(-1.0, 1.0)
+    # )
+    # counterfactual_data.X = Float32.(counterfactual_data.X)
+    return (X, y)
+end