Merge pull request #27 from JuliaTrustworthyAI/issue-with-seed

couple more things

CHANGELOG.md

@@ -10,4 +10,4 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 
 ### Changed
 
-- Changed the way the default seed is set to avoid overriding the global seed. 
+- Changed the way the default seed is set to avoid overriding the global seed. [#26], [#27]

src/TaijaData.jl

@@ -11,7 +11,13 @@ using Flux
 using MLDatasets
 using StatsBase
 
+"""
+    data_seed
+
+A global seed to produce standardized synthetic data. This seed is used in various functions to ensure reproducibility of the synthetic datasets.
+"""
 const data_seed = 42
+
 data_dir = joinpath(artifact"data-tabular", "data-tabular")
 
 include("synthetic/blobs.jl")

src/synthetic/linearly_separable.jl

@@ -2,8 +2,11 @@
     load_linearly_separable(n=250; seed=data_seed)
 
 Loads linearly separable synthetic data.
+
+!!! note
+    This calls the [`load_blobs`](@ref) function with specific parameters and the seed set to [`data_seed`](@ref). To ensure linear spearability and reproducibility, setting the `seed` keyword argument has no effect. For more flexibility, you can use [`load_blobs`](@ref) directly with different parameters if needed.
 """
 function load_linearly_separable(n=250; seed=data_seed)
-    data = load_blobs(n; seed=seed, centers=2, cluster_std=0.5)
+    data = load_blobs(n; seed=data_seed, centers=2, cluster_std=0.5)
     return data
 end

src/synthetic/moons.jl

@@ -2,6 +2,8 @@
     load_moons(n=250; seed=data_seed, kwrgs...)
 
 Loads synthetic moons data.
+
+
 """
 function load_moons(n=250; seed=data_seed, kwrgs...)
     if isa(seed, Int)

src/synthetic/multi_class.jl

@@ -1,7 +1,10 @@
 """
-    load_multi_class(n=250; seed=data_seed)
+    load_multi_class(n=250; seed=data_seed, centers=4)
 
 Loads multi-class synthetic data.
+
+!!! note
+    This calls the [`load_blobs`](@ref) function with specific parameters and the seed set to [`data_seed`](@ref).
 """
 function load_multi_class(n=250; seed=data_seed, centers=4)
     data = load_blobs(n; seed=seed, centers=centers, cluster_std=0.5)

src/synthetic/overlapping.jl

@@ -2,8 +2,11 @@
     load_overlapping(n=250; seed=data_seed)
 
 Loads overlapping synthetic data.
+
+!!! note
+    This calls the [`load_blobs`](@ref) function with specific parameters and the seed set to [`data_seed`](@ref). To ensure overlapping clusters and reproducibility, setting the `seed` keyword argument has no effect. For more flexibility, you can use [`load_blobs`](@ref) directly with different parameters if needed.
 """
 function load_overlapping(n=250; seed=data_seed)
-    data = load_blobs(n; seed=seed, centers=2, cluster_std=2.0)
+    data = load_blobs(n; seed=data_seed, centers=2, cluster_std=2.0)
     return data
 end