Merge pull request #76 from JuliaML/cl/tu

CarloLucibello · web-flow · commit f33886eb9699 · 2021-09-06T10:09:01.000+02:00
add TUDataset
diff --git a/README.md b/README.md
@@ -27,12 +27,10 @@ Find below a list of available datasets and links to their documentation.
   - [MNIST](https://juliaml.github.io/MLDatasets.jl/dev/datasets/MNIST/)
   - [SVHN2](https://juliaml.github.io/MLDatasets.jl/dev/datasets/SVHN2/)
 
-
 #### Miscellaneous
   - [BostonHousing](https://juliaml.github.io/MLDatasets.jl/dev/datasets/BostonHousing/)
   - [Iris](https://juliaml.github.io/MLDatasets.jl/dev/datasets/Iris/)
 
-
 #### Text
   - [PTBLM](https://juliaml.github.io/MLDatasets.jl/dev/datasets/PTBLM/)
   - [UD_English](https://juliaml.github.io/MLDatasets.jl/dev/datasets/UD_English/)
@@ -41,7 +39,7 @@ Find below a list of available datasets and links to their documentation.
   - [CiteSeer](https://juliaml.github.io/MLDatasets.jl/dev/datasets/CiteSeer/)
   - [Cora](https://juliaml.github.io/MLDatasets.jl/dev/datasets/Cora/)
   - [PubMed](https://juliaml.github.io/MLDatasets.jl/dev/datasets/PubMed/)
-
+  - [TUDatasets](https://juliaml.github.io/MLDatasets.jl/dev/datasets/TUDataset/)
 
 
 ## Installation
diff --git a/docs/make.jl b/docs/make.jl
@@ -44,6 +44,7 @@ makedocs(
                 "CiteSeer" => "datasets/CiteSeer.md",
                 "Cora" => "datasets/Cora.md",
                 "PubMed" => "datasets/PubMed.md",
+                "TUDataset" => "datasets/TUDataset.md",      
             ],
 
         ],
diff --git a/docs/src/datasets/SVHN2.md b/docs/src/datasets/SVHN2.md
@@ -128,7 +128,7 @@ SVHN2.classnames
 SVHN2.convert2image
 ```
 
-## References
+### References
 
 - **Authors**: Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bissacco, Bo Wu, Andrew Y. Ng
 
diff --git a/docs/src/datasets/TUDataset.md b/docs/src/datasets/TUDataset.md
@@ -0,0 +1,5 @@
+# TUDataset
+
+```@docs
+TUDataset
+```
diff --git a/docs/src/datasets/UD_English.md b/docs/src/datasets/UD_English.md
@@ -11,7 +11,7 @@ devdata = UD_English.devdata()
 testdata = UD_English.devdata()
 ```
 
-## Data Size
+# Data Size
 
 |    | Train x | Train y | Test x | Test y |
 |:--:|:-------:|:-------:|:------:|:------:|
diff --git a/src/BostonHousing/BostonHousing.jl b/src/BostonHousing/BostonHousing.jl
@@ -44,7 +44,7 @@ Harrison and Rubinfeld do not mention any censoring.
 The data file stored in this repo is a copy of the This is a copy of UCI ML housing dataset. 
 https://archive.ics.uci.edu/ml/machine-learning-databases/housing/
 
-## Interface
+# Interface
 
 - [`BostonHousing.features`](@ref)
 - [`BostonHousing.targets`](@ref)
diff --git a/src/CiteSeer/CiteSeer.jl b/src/CiteSeer/CiteSeer.jl
@@ -10,11 +10,11 @@ The dataset is designed for the node classification task.
 The task is to predict the category of certain paper.
 The dataset is retrieved from Ref. [2].
 
-## Interface
+# Interface
 
 - [`CiteSeer.dataset`](@ref)
 
-## References
+# References
 
 [1]: [Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking](https://arxiv.org/abs/1707.03815)
 [2]: [Planetoid](https://github.com/kimiyoung/planetoid)
@@ -63,7 +63,7 @@ the reverse of each edge and the graph will be undirected.
 
 See also [`CiteSeer`](@ref).
 
-## Usage Examples
+# Usage Examples
 
 ```julia
 using MLDatasets: CiteSeer
diff --git a/src/Cora/Cora.jl b/src/Cora/Cora.jl
@@ -11,7 +11,7 @@ The dataset is designed for the node classification task.
 The task is to predict the category of certain paper.
 The dataset is retrieved from Ref. [2].
 
-## Statistics 
+# Statistics 
 
 - Nodes: 2708
 - Edges: 10556
@@ -24,11 +24,11 @@ The dataset is retrieved from Ref. [2].
 The split is the one used in the original paper [1] and 
 doesn't consider all nodes.
 
-## Interface
+# Interface
 
 - [`Cora.dataset`](@ref)
 
-## References
+# References
 
 [1]: [Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking](https://arxiv.org/abs/1707.03815)
 [2]: [Planetoid](https://github.com/kimiyoung/planetoid
@@ -78,7 +78,7 @@ the reverse of each edge and the graph will be undirected.
 
 See also [`Cora`](@ref).
 
-## Usage Examples
+# Usage Examples
 
 ```juliarepl
 using MLDatasets: Cora
diff --git a/src/FashionMNIST/FashionMNIST.jl b/src/FashionMNIST/FashionMNIST.jl
@@ -12,12 +12,12 @@ examples. Each example is a 28x28 grayscale image, associated
 with a label from 10 classes. It can serve as a drop-in
 replacement for MNIST.
 
-## Interface
+# Interface
 
 - [`FashionMNIST.traintensor`](@ref), [`FashionMNIST.trainlabels`](@ref), [`FashionMNIST.traindata`](@ref)
 - [`FashionMNIST.testtensor`](@ref), [`FashionMNIST.testlabels`](@ref), [`FashionMNIST.testdata`](@ref)
 
-## Utilities
+# Utilities
 
 - [`FashionMNIST.download`](@ref)
 
diff --git a/src/Iris/Iris.jl b/src/Iris/Iris.jl
@@ -13,12 +13,12 @@ The module retrieves the data from the [UCI Machine Learning Repository](https:/
 
 NOTE: no pre-defined train-test split for this dataset, `features` and `labels` return the whole dataset. 
 
-## Interface
+# Interface
 
 - [`Iris.features`](@ref)
 - [`Iris.labels`](@ref)
 
-## Utilities
+# Utilities
 
 - [`Iris.download`](@ref)
 """
diff --git a/src/MLDatasets.jl b/src/MLDatasets.jl
@@ -1,10 +1,17 @@
 module MLDatasets
 
+using FixedPointNumbers: length
+using ColorTypes: length
 using Requires
 using DelimitedFiles: readdlm
 using FixedPointNumbers, ColorTypes
 using PyCall
 
+# Julia 1.0 compatibility
+if !isdefined(Base, :isnothing)
+    isnothing(x) = x === nothing
+end
+
 bytes_to_type(::Type{UInt8}, A::Array{UInt8}) = A
 bytes_to_type(::Type{N0f8}, A::Array{UInt8}) = reinterpret(N0f8, A)
 bytes_to_type(::Type{T}, A::Array{UInt8}) where T<:Integer = convert(Array{T}, A)
@@ -51,6 +58,7 @@ include("planetoid.jl")
     include("Cora/Cora.jl")
     include("PubMed/PubMed.jl")
     include("CiteSeer/CiteSeer.jl")
+include("TUDataset/TUDataset.jl")
 
 function __init__()
     # initialize optional dependencies
@@ -71,6 +79,8 @@ function __init__()
         out = out.todense() if hasattr(out, 'todense') else out
         return out
     """
+
+    __init__tudataset()
 end
 
 end
diff --git a/src/MNIST/MNIST.jl b/src/MNIST/MNIST.jl
@@ -12,12 +12,12 @@ used in small-scale machine learning experiments. It contains
 pixel gray-scale image that depicts a handwritten version of 1 of
 the 10 possible digits (0-9).
 
-## Interface
+# Interface
 
 - [`MNIST.traintensor`](@ref), [`MNIST.trainlabels`](@ref), [`MNIST.traindata`](@ref)
 - [`MNIST.testtensor`](@ref), [`MNIST.testlabels`](@ref), [`MNIST.testdata`](@ref)
 
-## Utilities
+# Utilities
 
 - [`MNIST.download`](@ref)
 - [`MNIST.convert2image`](@ref)
diff --git a/src/PubMed/PubMed.jl b/src/PubMed/PubMed.jl
@@ -10,11 +10,11 @@ The dataset is designed for the node classification task.
 The task is to predict the category of certain paper.
 The dataset is retrieved from Ref. [2].
 
-## Interface
+# Interface
 
 - [`PubMed.dataset`](@ref)
 
-## References
+# References
 
 [1]: [Deep Gaussian Embedding of Graphs: Unsupervised Inductive Learning via Ranking](https://arxiv.org/abs/1707.03815)
 [2]: [Planetoid](https://github.com/kimiyoung/planetoid)
@@ -63,7 +63,7 @@ the reverse of each edge and the graph will be undirected.
 
 See also [`PubMed`](@ref).
 
-## Usage Examples
+# Usage Examples
 
 ```julia
 using MLDatasets: PubMed
diff --git a/src/SVHN2/SVHN2.jl b/src/SVHN2/SVHN2.jl
@@ -15,13 +15,13 @@ RGB color channel. The dataset is split up into three subsets:
 73257 digits for training, 26032 digits for testing, and 531131
 additional to use as extra training data.
 
-## Interface
+# Interface
 
 - [`SVHN2.traintensor`](@ref), [`SVHN2.trainlabels`](@ref), [`SVHN2.traindata`](@ref)
 - [`SVHN2.testtensor`](@ref), [`SVHN2.testlabels`](@ref), [`SVHN2.testdata`](@ref)
 - [`SVHN2.extratensor`](@ref), [`SVHN2.extralabels`](@ref), [`SVHN2.extradata`](@ref)
 
-## Utilities
+# Utilities
 
 - [`SVHN2.download`](@ref)
 - [`SVHN2.classnames`](@ref)
diff --git a/src/TUDataset/TUDataset.jl b/src/TUDataset/TUDataset.jl
diff --git a/test/runtests.jl b/test/runtests.jl
diff --git a/test/tst_tudataset.jl b/test/tst_tudataset.jl
