search

hurlbertlab / core-transient

Data and code for NSF funded research on core vs transient species
7 stars 3 forks source link

histograms and beta distributions #18

Closed bsevansunc closed 9 years ago

bsevansunc commented 9 years ago

I've updated the histogram plots. The beta distribution is now shown between 1/nTime and 1-1/nTime where nTime is the number of time intervals in the sample. I've noticed that the beta distribution doesn't seem to be capturing the distribution of the data on the left side of the plot (towards transients). Prior to making more of these, I'm going to move onto to making the summary plots across sites (grouping by system and taxonomic class) but I was wondering if you guys have any suggestions regarding these beta distributions?

ahhurlbert commented 9 years ago

I don't think 1-1/nTime should be the upper limit of the interval over which beta is shown. For example, see p. 494. The line should extend to the equivalent of halfway through the bar representing 1.0. But separate from how much of it we should see, perhaps we need to revisit how it's being fit (e.g. is it being fit to some scaled data, and then rescaled back; are these choices appropriate, etc...).

bsevansunc commented 9 years ago

Sounds good, will look into it. Also, I put more plots up in the plots folder, let me know what you think and I'll adjust as needed. On Jan 3, 2015 3:00 PM, "ahhurlbert" notifications@github.com wrote:

I don't think 1-1/nTime should be the upper limit of the interval over which beta is shown. For example, see p. 494. The line should extend to the equivalent of halfway through the bar representing 1.0. But separate from how much of it we should see, perhaps we need to revisit how it's being fit (e.g. is it being fit to some scaled data, and then rescaled back; are these choices appropriate, etc...).

— Reply to this email directly or view it on GitHub https://github.com/hurlbertlab/core-transient/issues/18#issuecomment-68607459 .

ahhurlbert commented 9 years ago

They look good. I'm thinking an alternative to the plots showing average percent core and transient by system and taxa would be a stacked bar showing core, neither, and transient in that order. This means we lose the ability to easily show variation within a group (i.e. the error bars), but it might help make it easier to see the extent to which species truly tend to be in one of the extreme modes versus the neither category.