source: prdatasets/musk2.m

Last change on this file was 150, checked in by bduin, 5 years ago
File size: 1.2 KB
Line 
1%MUSK2 Prediction of molecules to be musks or non-musks
2%PRTools dataset import from UCI, 6598 objects, 166 features, 2 classes
3%
4%  A = MUSK2
5%
6%DESCRIPTION
7%This command downloads one of the UCI data sets, converts it into PRTools
8%format and stores it locally for future use. Consult the <a href="http://archive.ics.uci.edu/ml/datasets/Musk+(Version+2)">related website</a>
9%for further information. Please make the appropriate references in
10%publications that make use of this dataset.
11%
12%A is a multiple-instance dataset. The objects are the instances. They
13%group in 102 bags. There are two labels: (1) the Bag Label and (2) the Bag
14%Index. Use the CHANGELABLIST command to change from labeling system. See
15%also MULTI_LABELING
16%
17%SEE ALSO <a href="http://prtools.tudelft.nl/prtools/">PRTools Guide</a>, <a href="http://archive.ics.uci.edu/ml/">UCI Website</a>
18%DATASETS, MULTI_LABELING, CHANGELABLIST
19
20function a = musk2
21
22% UCI data is Z-compressed. No Matlab command foud for uncompressing.
23% Load zipped data from prtools website.
24a = pr_download_uci('Musk+(Version+2)','http://prtools.tudelft.nl/prdatasets/musk2.zip',[],[],5);
25a = setlablistnames(a,'Bag Label',1);
26a = feat2lab(a,1,'Bag Index');
27a(:,2) = [];
28a = setname(a,'Musk version 2');
Note: See TracBrowser for help on using the repository browser.