readme.md

October 26, 2016 ยท View on GitHub

Experiment Data

DataTaskLink#Train_Set#FeatureComments
HiggsBinary classificationlink10,500,00028use last 500,000 samples as test set
Yahoo LTRLearning to ranklink473,134700set1.train as train, set1.test as test
MS LTRLearning to ranklink2,270,296137{S1,S2,S3} as train set, {S5} as test set

How to generate data

  1. download above data set
  2. move "HIGGS.CSV" in Higgs data set to this folder, and run "higgs2libsvm.py"
  3. move "set1.train.txt" and "set1.test.txt" in Yahoo data set to this folder, and run "yahoo2libsvm.py"
  4. move "train.txt" and "test.txt" in Fold1 of MSLTR data set to this folder, and run "msltr2libsvm.py"