I come from the field of Numerical Analysis, and I look for the term which describes the problem of fitting a probability distribution to statistical numerical continuous data, without a-priori knowledge of the distribution.

Specifically, I read about nonparametric-statistics and in a lot of places it says that it best fit ordinal data (voting, categories, labels etc.), and not numerical.

What is the proper term I should be reading about? Am I wrong in this characterization of non-parametric statistics?

I should remark that I don't aim for the most precise definition, but rather the one that would allow me to contact people, ask questions and look for papers and textbooks.

  • 1
    $\begingroup$ You don't actually define nonparametric statistics anywhere that I see, so your question about being wrong in your definition seems odd -- but in truth many nonparametric methods are explicitly designed for numerical data and may require some adapting to use with ordinal data. If you want to learn about nonparametric stats, maybe start with a text. Conover's book (practical nonparametric statistics) - though it has been around a long time - is probably worth a look. $\endgroup$
    – Glen_b
    Aug 10, 2016 at 4:28
  • $\begingroup$ Thanks Glen. Are there any specific basic statistical concepts I should study before aiming at it? $\endgroup$
    – Amir Sagiv
    Aug 10, 2016 at 5:54
  • 1
    $\begingroup$ I don't know what you have already. It doesn't really require much heavy machinery, but you need some probability, and a good handle on permutations and combinations and a little algebra. Some knowledge of the basics of hypothesis testing would be very useful. $\endgroup$
    – Glen_b
    Aug 10, 2016 at 6:12
  • 1
    $\begingroup$ Look into kernel density estimation $\endgroup$ Jan 16, 2019 at 15:12


Your Answer

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

Browse other questions tagged or ask your own question.