search

biocore / evident

BSD 3-Clause "New" or "Revised" License
14 stars 9 forks source link

non-pairwise effect sizes #30

Open melissaszy opened 10 months ago

melissaszy commented 10 months ago

Hello,

I trying to use evident and am wondering if there is a way to filter the data before calculating non-pairwise effect sizes.

I want to calculate the non-pairwise effect size of one category (species). I have five variables currently in that category (water, dog, cat, rabbit, mouse). Is it possible to calculate non-pairwise effect size of only dog, cat, rabbit, mouse ie by first filtering out the water samples?

Alternatively, is it possible to derive non-pairwise effect sizes for the four species based on pairwise effect sizes? For instance, I calculate pairwise effect sizes based on all five variables (water, dog, cat, rabbit, mouse) then I ignore the pairwise effect sizes that involves water samples. Using the remaining effect sizes (related to dog, cat, rabbit, mouse), can I calculate a non-pairwise effect size?

Thank you for your time.

antgonza commented 10 months ago

Hi @melissaszy, if I understand correctly your question: yes, you can simply ignore the pairwise calculation against water. However, I think it will be best if you remove those samples from your feature table or distance matrix and then calculate the effect size of the rest of the samples. Hope this helps.

melissaszy commented 9 months ago

Thank you for your help!