My apologies if this question is trivial but I am having a problem to interpret the Bk plot from dendextend. There are 3 lines there: dented one to show FM index for every k, then red solid one and dashed black one. What are those last two lines? I thought that they represents the critical value of FM index under H0 but the values not sems to match... Could you help please?
Asked
Active
Viewed 129 times
1 Answers
1
The middle value indeed is the expected FM value under H0, while the other two lines are the approximate confidence intervals of that value. You can see the original paper for more details: http://wildfire.stat.ucla.edu/pdflibrary/fowlkes.pdf
Tal Galili
- 24,605
- 44
- 129
- 187
-
Thank you for getting back to me!!!! Dendextend is a GREAT package btw!!!!! I think in the original paper there were 3 lines superimposed with main plot. E(X) (expected X value, in this case X=Bk) and E(Bk)+/-2sqrt(var(Bk)). But I thought they are all sort of obsolete. Dont you think that the plot would benefit more from the line (sort of line, it would be connected points) representing critical values under H0? It will allow quick check whether there is certain number of clusters k which is resulting in lack of grounds to accept H0 that the trees are not-smilar... – AussieAndy Feb 16 '18 at 15:07
-
Hi Aussie, good question. I wanted to be consistent with the original paper, and thought that would be a good way of doing it.If you'd like to go over the functions and add an argument to have what you suggest, I'd be willing to review the code and add it to the repo :) Cheers, T – Tal Galili Feb 17 '18 at 08:36
-
Hi Tal - it would be GREAT if the plot can be altered this way! I really think it will be more informative! Thank you in advance :) – AussieAndy Feb 17 '18 at 09:25
-
Hi, I opened an issue for this, but I'm not likely to get to do it in the near future. If you get to write something, feel free to respond on that issue: https://github.com/talgalili/dendextend/issues/68 – Tal Galili Feb 17 '18 at 13:46