0
$\begingroup$

I would like to compare two datasets. Each dataset includes the ratio of chosen features (number of chosen tokens/number of all tokens in the document). The distribution is not normal.

I want to check of one group uses the features more often than the other one.

Would the Mann-Whitney U Test be suitable for me in this case? Which alternative should I take (SciPy suggests two-sided, less, and greater alternative)?

Improve this question
$\endgroup$
3
  • 2
    $\begingroup$ What question(s) do you have of your data? $\endgroup$
    – Dave
    Commented Jul 19, 2021 at 17:13
  • 1
    $\begingroup$ Do you want to estimate stochastic dominance between the two datasets using MWU test? Related: stats.stackexchange.com/questions/533424/… $\endgroup$
    – Galen
    Commented Jul 19, 2021 at 17:21
  • $\begingroup$ I want to check if one group uses selected features more often than the other one $\endgroup$
    – dar1234
    Commented Jul 19, 2021 at 17:45

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.