zum Inhalt springen

Data for Classification

This site contains data sets used in the joint project of the University of Cologne and the Hochschule Merseburg “Classifying real-world data with the DD?-procedure”. Comprehensive description of the methodology, and experimental settings and results of the study are presented in the work:

Mozharovskyi, P., Mosler, K. and Lange, T. (2014): Classifying real-world data with the DD?-procedure. Advances in Data Analysis and Classification, to appear.

For a more complete explanation of the technique and further experiments see:
Lange, T., Mosler, K. and Mozharovskyi, P. (2012): “Fast nonparametric classification based on data depth”. Statistical Papers 55(1), 49-69. (The final publication is available at www.springerlink.com).

50 binary classification tasks have been obtained from partitioning 33 freely accessible data sets. Multiclass problems were reasonably split into binary classification problems, some of the data set were slightly processed by removing objects or attributes and selecting prevailing classes. Each data set is provided with a (short) description and brief descriptive statistics. The name reflects the origination of the data. A letter after the name is a property filter, letters (also their combinations) in brackets separated by "vs" are the classes opposed. The letters (combinations or words) stand for labels of classes (names of properties) and are intuitive. Each description contains a link to the original data.

The data have been collected as open source data in January 2013. Owners of this web page decline any responsibility regarding their correctness or consequences of their usage. If you publish material based on these data, please quote the original source. Special requests regarding citations are found on data set's web page.

The general list of sources consists of:

http://archive.ics.uci.edu/ml, se also Frank, A. & Asuncion, A. (2010). UCI Machine Learning Repository [http://archive.ics.uci.edu/ml]. Irvine, CA: University of California, School of Information and Computer Science.

All the data sets as *.zip:  

Data table

#Datasetn1n2n1+n2dln(n1/n2)(n1+n2)/d# tiedDownload
4.Blood Transfusion1785707483-1,171249,3246
5.Breast Cancer Wisconsin45824169990,64277,7236
6.Bupa Liver Disorder1452003456-0,32957,54
7.Chemical Diabetes (C vs N)36761125-0,75522,40
8.Chemical Diabetes (C vs O)36336950,08613,80
9.Chemical Diabetes (N vs O)763310950,83321,80
11.Crabs (B vs O)1001002005040,00
12.Crabs (M vs F)1001002005040,00
13.Crabs B (M vs F)50501005020,00
14.Crabs F (B vs O)50501005020,00
15.Crabs M (B vs O)50501005020,00
16.Crabs O (M vs F)50501005020,00
17.Cricket (C vs P)78781564039,07
18.Diabetes (of Pima Indians)2685007688-0,61696,00
19.Ecoli (CP vs IM)1437722050,62144,00
20.Ecoli (CP vs PP)1435219551,01239,00
21.Ecoli (IM vs PP)775212950,39225,80
22.Gemsen (M vs F)796553134960,365224,827
23.Glass (F vs NF)70761469-0,08316,21
24.Groessen (M vs F)11611423030,02076,70
25.Haberman's Survival2258130631,022102,023
28.Indian Liver Patient (1 vs 2)414165579100,92057,913
29.Indian Liver Patient (M vs F)1404395799-1,13964,313
30.Iris Plants (SET vs VER)50501004025,02
31.Iris Plants (SET vs VIR)50501004025,03
32.Iris Plants (VER vs VIR)50501004025,01
33.Irish Educational Transitions (M vs F)25025050050100,044
34.Kidney (M vs F)2056765-1,02215,20
35.PIMA (training)1326820070,66328,60
36.Plasma Retinol and Beta-Carotene Levels (M vs F)27342315131,87224,20
37.Segmentation (C vs W)33033066010066,062
38.Social Mobility (I vs NI)578578115650231,245
39.Social Mobility (W vs B)578578115650231,28
40.Teaching Assistan Evaluation (E vs NE)291221515-1,42730,243
41.Tennis (M vs F)42458715-0,0735,80
42.Tips (D vs N)1766824460,95240,71
43.Tips (M vs F)871572446-0,59840,71
44.US Crime (S vs N)16314713-0,6543,60
45.Vertebral Column21010031060,74251,70
46.Veteran Lung Cancer (S vs T)696813770,01019,60
47.Vowel (M vs F)528462990130,13176,20
48.Wine (1 vs 2)597113013-0,18610,00
49.Wine (1 vs 3)5948107130,2078,20
50.Wine (2 vs 3)7148119130,3929,20