0

In the examples given at this link, I am not able to decide whether the distribution is unimodal or bimodal. I think it is in between unimodal and bimodal, but I do not know if this kind of class exists or not.

Can anyone please suggest something so that I can easily understand in which case I should say unimodal or bimodal distribution!

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
Praveen Mishra
  • 9
  • 1
  • 2
    That link contains no data or examples and the links from it do not work. Please endeavor to make your question stand by itself by including (a representative subset of) the data in the question and/or linking to an image. (If the link is legitimate, community members can embed it into the question itself on your behalf.) When you do this, could you please also add a few words explaining *why* you want to determine the number of modes. What conclusions will you draw from that? – whuber May 15 '13 at 13:15
  • One thing you could try, which is not necessarly the best one, is to compute the modes of your distribution and analyse a) if they are close/far to each other in terms b) scale of the density at diferent modes (different peaks of density when they exist) you would have selected (you want to eleminate the node when value of density at this point is very close to zero) . Exampe : you may select 3 modes of the $N(0,1)$ which are $-\infty , 0 , + \infty$ and you decide it is mono-modal – dfhgfh May 15 '13 at 13:31
  • 1
    There is a test of bimodality called the "dip test", developed by Hartigan, that might help. – Peter Flom May 15 '13 at 13:33
  • The links and references and discussion [here](http://stats.stackexchange.com/a/51085/805) may be of some help. – Glen_b May 16 '13 at 02:11

0 Answers0