LIBSVMdata.jl
A simple tool to fetch LIBSVM datasets to Julia.
Some datasets are not covered by CI tests due to their data size. Please report any bug in the issues page.
Installation
LIBSVMdata
can be installed as follows :
pkg> add LIBSVMdata
Usage
The design matrix A
and the response variable y
corresponding to a given dataset (a1a
in this example) can be loaded as :
using LIBSVMdata
A, y = load_dataset("a1a",
dense = false,
replace = false,
verbose = true,
)
All the datasets are stored in the folder specified by the environnement variable LIBSVMDATA_HOME
if it exists. Otherwise, they are stored under $HOME/data/libsvm
.
Arguments :
dataset::String
: Name of the dataset. All the available datasets can be found using theprint_datasets()
function.dense::Bool=false
: Iftrue
, the feature matrixA
returned is dense. Otherwise, it is sparse.replace::Bool=false
: Iftrue
, the dataset is re-downloaded even if it already exists in the dataset directory.verbose::Bool=true
: Iftrue
, toggle verbosity.
Available datasets
The available datasets can be listed using the print_datasets()
function. This function also specifies the dataset type, the number of data points, the number of features and the number of different classes. If the dataset is a regression one, the number of classes is arbitrarily set to Inf
.
using LIBSVMdata
print_datasets()
==================================================================================
Dataset name | Type | Data | Features | Classes
----------------------------------------------------------------------------------
a1a | binary | 1605 | 123 | 2
a1a.t | binary | 30956 | 123 | 2
a2a | binary | 2265 | 123 | 2
a2a.t | binary | 30296 | 123 | 2
a3a | binary | 3185 | 123 | 2
a3a.t | binary | 29376 | 123 | 2
a4a | binary | 4781 | 123 | 2
a4a.t | binary | 27780 | 123 | 2
a5a | binary | 6414 | 123 | 2
a5a.t | binary | 26147 | 123 | 2
a6a | binary | 11220 | 123 | 2
a6a.t | binary | 21341 | 123 | 2
a7a | binary | 16100 | 123 | 2
a7a.t | binary | 16461 | 123 | 2
a8a | binary | 22696 | 123 | 2
a8a.t | binary | 9865 | 123 | 2
a9a | binary | 32561 | 123 | 2
a9a.t | binary | 16281 | 123 | 2
australian | binary | 690 | 14 | 2
australian_scale | binary | 690 | 14 | 2
avazu-app | binary | 14596137 | 1000000 | 2
avazu-app.t | binary | 1719304 | 1000000 | 2
avazu-app.tr | binary | 12642186 | 1000000 | 2
avazu-app.val | binary | 1953951 | 1000000 | 2
avazu-site | binary | 25831830 | 1000000 | 2
avazu-site.t | binary | 2858160 | 1000000 | 2
avazu-site.tr | binary | 23567843 | 1000000 | 2
avazu-site.val | binary | 2264987 | 1000000 | 2
breast-cancer | binary | 683 | 10 | 2
breast-cancer_scale | binary | 683 | 10 | 2
cod-rna | binary | 59535 | 8 | 2
cod-rna.t | binary | 271617 | 8 | 2
cod-rna.r | binary | 157413 | 8 | 2
colon-cancer | binary | 62 | 2000 | 2
covtype.binary | binary | 581012 | 54 | 2
covtype.binary.scale | binary | 581012 | 54 | 2
diabetes | binary | 768 | 8 | 2
diabetes_scale | binary | 768 | 8 | 2
duke | binary | 44 | 7129 | 2
duke.tr | binary | 38 | 7129 | 2
duke.val | binary | 4 | 7129 | 2
epsilon | binary | 400000 | 2000 | 2
epsilon.t | binary | 100000 | 2000 | 2
fourclass | binary | 862 | 2 | 2
fourclass_scale | binary | 862 | 2 | 2
german.number | binary | 1000 | 24 | 2
german.number_scale | binary | 1000 | 24 | 2
gisette_scale | binary | 6000 | 5000 | 2
gisette_scale.t | binary | 1000 | 5000 | 2
heart | binary | 270 | 13 | 2
heart_scale | binary | 270 | 13 | 2
HIGGS | binary | 11000000 | 28 | 2
ijcnn | binary | 49990 | 22 | 2
ijcnn.t | binary | 91701 | 22 | 2
ijcnn.tr | binary | 35000 | 22 | 2
ijcnn.val | binary | 14990 | 22 | 2
ionosphere_scale | binary | 351 | 34 | 2
kdda | binary | 8407752 | 20216830 | 2
kdda.t | binary | 510302 | 20216830 | 2
kddb | binary | 19264097 | 29890095 | 2
kddb.t | binary | 748401 | 29890095 | 2
kdd12 | binary | 149639105 | 54686452 | 2
kdd12.tr | binary | 119705032 | 54686452 | 2
kdd12.val | binary | 29934073 | 54686452 | 2
leukemia | binary | 38 | 7129 | 2
leukemia.t | binary | 34 | 7129 | 2
liver-disorders | binary | 145 | 5 | 2
liver-disorders_scale | binary | 145 | 5 | 2
liver-disorders.t | binary | 200 | 5 | 2
madelon | binary | 2000 | 500 | 2
madelon.t | binary | 600 | 500 | 2
mushrooms | binary | 8124 | 112 | 2
news20.binary | binary | 19996 | 1355191 | 2
phishing | binary | 11055 | 68 | 2
rcv1_train.binary | binary | 20242 | 47236 | 2
rcv1_test.binary | binary | 677399 | 47236 | 2
ream-sim | binary | 72309 | 20958 | 2
skin_nonskin | binary | 245057 | 3 | 2
splice | binary | 1000 | 60 | 2
splice_scale | binary | 1000 | 60 | 2
splice.t | binary | 2175 | 60 | 2
splice-site | binary | 50000000 | 11725480 | 2
splice-site.t | binary | 4627840 | 11725480 | 2
sonar_scale | binary | 208 | 60 | 2
SUSY | binary | 5000000 | 18 | 2
svmguide1 | binary | 3089 | 4 | 2
svmguide1.t | binary | 4000 | 4 | 2
svmguide3 | binary | 1243 | 22 | 2
svmguide3.t | binary | 41 | 22 | 2
url | binary | 2396130 | 3231961 | 2
url_normalized | binary | 2396130 | 3231961 | 2
w1a | binary | 2477 | 300 | 2
w1a.t | binary | 47272 | 300 | 2
w2a | binary | 3470 | 300 | 2
w2a.t | binary | 46279 | 300 | 2
w3a | binary | 4912 | 300 | 2
w3a.t | binary | 44837 | 300 | 2
w4a | binary | 7366 | 300 | 2
w4a.t | binary | 42383 | 300 | 2
w5a | binary | 9888 | 300 | 2
w5a.t | binary | 39861 | 300 | 2
w6a | binary | 17188 | 300 | 2
w6a.t | binary | 32561 | 300 | 2
w7a | binary | 24692 | 300 | 2
w7a.t | binary | 25057 | 300 | 2
w8a | binary | 49749 | 300 | 2
w8a.t | binary | 14951 | 300 | 2
webspam-trigram | binary | 350000 | 16609143 | 2
webspam-unigram | binary | 350000 | 16609143 | 2
aloi | multiclass | 108000 | 128 | 1000
aloi.scale | multiclass | 108000 | 128 | 1000
cifar10 | multiclass | 50000 | 3072 | 10
cifar10.t | multiclass | 50000 | 3072 | 10
connect-4 | multiclass | 67557 | 126 | 3
covtype | multiclass | 581012 | 54 | 7
covtype.scale01 | multiclass | 581012 | 54 | 7
covtype.scale | multiclass | 581012 | 54 | 7
dna.scale | multiclass | 2000 | 180 | 3
dna.scale.t | multiclass | 1186 | 180 | 3
dna.scale.tr | multiclass | 1400 | 180 | 3
dna.scale.val | multiclass | 600 | 180 | 3
glass.scale | multiclass | 214 | 9 | 6
iris.scale | multiclass | 150 | 4 | 3
letter.scale | multiclass | 15000 | 16 | 26
letter.scale.t | multiclass | 5000 | 16 | 26
letter.scale.tr | multiclass | 10500 | 16 | 26
letter.scale.val | multiclass | 4500 | 16 | 26
mnist | multiclass | 60000 | 780 | 10
mnist.t | multiclass | 10000 | 778 | 10
mnist.scale | multiclass | 60000 | 780 | 10
mnist.scale.t | multiclass | 10000 | 778 | 10
mnist8m | multiclass | 8100000 | 784 | 10
mnist8m.scale | multiclass | 8100000 | 784 | 10
news20 | multiclass | 15935 | 62061 | 20
news20.t | multiclass | 3993 | 62060 | 20
news20.scale | multiclass | 15935 | 62061 | 20
news20.t.scale | multiclass | 3993 | 62060 | 20
pendigits | multiclass | 7494 | 16 | 10
pendigits.t | multiclass | 3498 | 16 | 10
poker | multiclass | 25010 | 10 | 10
poker.t | multiclass | 1000000 | 10 | 10
protein | multiclass | 17766 | 357 | 3
protein.t | multiclass | 6621 | 357 | 3
protein.tr | multiclass | 14895 | 357 | 3
protein.val | multiclass | 2871 | 357 | 3
rcv1_tain.multiclass | multiclass | 15564 | 47236 | 53
rcv1_test.multiclass | multiclass | 518571 | 47236 | 53
satimage.scale | multiclass | 4435 | 36 | 6
satimage.scale.t | multiclass | 2000 | 36 | 6
satimage.scale.tr | multiclass | 3104 | 36 | 6
satimage.scale.val | multiclass | 1331 | 36 | 6
sector | multiclass | 6412 | 55197 | 105
sector.t | multiclass | 3207 | 55197 | 105
sector.scale | multiclass | 6412 | 55197 | 105
sector.scale.t | multiclass | 3207 | 55197 | 105
segment.scale | multiclass | 2310 | 19 | 7
Sensorless | multiclass | 58509 | 48 | 11
Sensorless.scale | multiclass | 58509 | 48 | 11
Sensorless.scale.tr | multiclass | 10000 | 48 | 11
shuttle.scale | multiclass | 43500 | 9 | 7
shuttle.scale.t | multiclass | 14500 | 9 | 7
shuttle.scale.tr | multiclass | 30450 | 9 | 7
shuttle.scale.val | multiclass | 13050 | 9 | 7
smallNORB | multiclass | 24300 | 18432 | 5
smallNORB.t | multiclass | 24300 | 18432 | 5
smallNORB-32x32 | multiclass | 24300 | 2048 | 5
smallNORB-32x32.t | multiclass | 24300 | 2048 | 5
SVHN | multiclass | 73257 | 3072 | 10
SVHN.t | multiclass | 26032 | 3072 | 10
SVHN.extra | multiclass | 531131 | 3072 | 10
SVHN.scale | multiclass | 73257 | 3072 | 10
SVHN.scale.t | multiclass | 26032 | 3072 | 10
SVHN.scale.extra | multiclass | 531131 | 3072 | 10
svmguide2 | multiclass | 391 | 20 | 3
svmguide4 | multiclass | 300 | 10 | 6
svmguide4.t | multiclass | 312 | 10 | 6
usps | multiclass | 7291 | 256 | 10
usps.t | multiclass | 2007 | 256 | 10
acoustic | multiclass | 78823 | 50 | 3
acoustic.t | multiclass | 19705 | 50 | 3
acoustic_scale | multiclass | 78823 | 50 | 3
acoustic_scale.t | multiclass | 19705 | 50 | 3
combined | multiclass | 78823 | 100 | 3
combined.t | multiclass | 19705 | 100 | 3
combined_scale | multiclass | 78823 | 100 | 3
combined_scale.t | multiclass | 19705 | 100 | 3
vehicle.scale | multiclass | 846 | 18 | 4
vowel | multiclass | 528 | 10 | 11
vowel.t | multiclass | 426 | 10 | 11
vowel.scale | multiclass | 528 | 10 | 11
vowel.scale.t | multiclass | 426 | 10 | 11
wine.scale | multiclass | 178 | 13 | 3
abalone | regression | 4177 | 8 | Inf
abalone_scale | regression | 4177 | 8 | Inf
bodyfat | regression | 252 | 14 | Inf
bodyfat_scale | regression | 252 | 14 | Inf
cadata | regression | 20640 | 8 | Inf
cpusmall | regression | 8192 | 12 | Inf
cpusmall_scale | regression | 8192 | 12 | Inf
E2006-log1p.train | regression | 16087 | 4272227 | Inf
E2006-log1p.test | regression | 3308 | 4272227 | Inf
E2006.train | regression | 16087 | 150360 | Inf
E2006.test | regression | 3308 | 150360 | Inf
eunite2001 | regression | 336 | 16 | Inf
eunite2001.t | regression | 31 | 16 | Inf
housing | regression | 506 | 13 | Inf
housing_scale | regression | 506 | 13 | Inf
mg | regression | 1385 | 6 | Inf
mg_scale | regression | 1385 | 6 | Inf
mgp | regression | 392 | 7 | Inf
mgp_scale | regression | 392 | 7 | Inf
pyrim | regression | 74 | 27 | Inf
space_ga | regression | 3107 | 6 | Inf
space_ga_scale | regression | 3107 | 6 | Inf
triazines | regression | 186 | 60 | Inf
triazines_scale | regression | 186 | 60 | Inf
YearPredictionMSD | regression | 463715 | 90 | Inf
YearPredictionMSD.t | regression | 51630 | 90 | Inf
bibtex | multilabel | 7395 | 1836 | 159
BlogCatalog_deepwalk | multilabel | 10312 | 128 | 39
BlogCatalog_line | multilabel | 10312 | 128 | 39
BlogCatalog_node2vec | multilabel | 10312 | 128 | 39
delicious | multilabel | 16105 | 500 | 983
eurlex_tfidf_train | multilabel | 15449 | 186104 | 3956
eurlex_tfidf_test | multilabel | 3865 | 186104 | 3956
flickr_deepwalk | multilabel | 80513 | 128 | 195
flickr_line | multilabel | 80513 | 128 | 195
flickr_node2vec | multilabel | 80513 | 128 | 195
exp1-train | multilabel | 30993 | 120 | 101
exp1-test | multilabel | 12914 | 120 | 101
ppi_deepwalk | multilabel | 56944 | 128 | 121
ppi_line | multilabel | 56944 | 128 | 121
ppi_node2vec | multilabel | 56944 | 128 | 121
rcv1subset_topics_train_1 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_train_2 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_train_3 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_train_4 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_train_5 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_test_1 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_test_2 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_test_3 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_test_4 | multilabel | 3000 | 47236 | 101
rcv1subset_topics_test_5 | multilabel | 3000 | 47236 | 101
rcv1_topics_train | multilabel | 23149 | 47236 | 101
rcv1_topics_test_0 | multilabel | 199328 | 47236 | 101
rcv1_topics_test_1 | multilabel | 199339 | 47236 | 101
rcv1_topics_test_2 | multilabel | 199576 | 47236 | 101
rcv1_topics_test_3 | multilabel | 183022 | 47236 | 101
rcv1_topics_combined_test | multilabel | 781265 | 47236 | 101
rcv1_industries_train | multilabel | 23149 | 47236 | 313
rcv1_industries_test_0 | multilabel | 23149 | 781265 | 313
rcv1_industries_test_1 | multilabel | 23149 | 781265 | 313
rcv1_industries_test_2 | multilabel | 23149 | 781265 | 313
rcv1_industries_test_3 | multilabel | 23149 | 781265 | 313
rcv1_regions_train | multilabel | 23149 | 47236 | 228
rcv1_regions_test_0 | multilabel | 23149 | 781265 | 228
rcv1_regions_test_1 | multilabel | 23149 | 781265 | 228
rcv1_regions_test_2 | multilabel | 23149 | 781265 | 228
rcv1_regions_test_3 | multilabel | 23149 | 781265 | 228
scene_train | multilabel | 1211 | 294 | 6
scene_test | multilabel | 1196 | 294 | 6
tmc2007_train | multilabel | 21519 | 30438 | 22
tmc2007_test | multilabel | 7077 | 30438 | 22
wiki10_31k_tfidf_train | multilabel | 14146 | 104374 | 30938
wiki10_31k_tfidf_test | multilabel | 6616 | 104374 | 30938
yeast_train | multilabel | 1500 | 103 | 14
yeast_test | multilabel | 917 | 103 | 14
youtube_deepwalk | multilabel | 1138499 | 128 | 46
youtube_line | multilabel | 1138499 | 128 | 46
youtube_node2vec | multilabel | 1138499 | 128 | 46
==================================================================================
Credits
The structure of this package is inspired from libsvmdata.