Machine Learning , Data Mining studies uses various data sets. If used data sets are not well known data sets, following points of these data sets should be explained. More can be added but these points comes to my mind immediately.

  1. features/attributes (number, data type, domain definition)

  2. How many instances exists in this data set?

  3. Test / Training data (Random Sampling, All data used..)

  4. Feature Reduction techniques

  5. Any preprocessing/post processing

Thus my question is: Where should I put this data set explanation in article sections?

  • Introduction
  • Methods (Experiments)
  • Or entirely new section under Data Set heading?
  • 4
    In the same place that the papers you cite did. – JeffE Feb 26 '14 at 1:34
  • In machine learning this is usually in materials and methods. – Marc Claesen Mar 16 '14 at 9:16

There is no universally-agreed way to do it. I usually place the information about the datasets under the experiments section, before reporting the results. Also, I prefer to present parameters about the datasets (e.g., number of features, number of samples) in a table. If many datasets are used, and there are some points that needs to be clarified, I would create a subsection for the datasets under the experiments section.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.