search

traitecoevo / modeladequacy

The purpose of this project is to develop an approach to evaluate the fit of continuous trait evolution models and to apply this approach to angiosperm functional trait data.
7 stars 2 forks source link

Weird datasets #83

Closed mwpennell closed 10 years ago

mwpennell commented 10 years ago

The problem i mentioned in issue #82 regarding some datasets failing has reared its head again. I have just run the bayesian analyses and the following clades got stuck and failed to produce output (29, 112, 119, 160, 161, 167, 203, 307). I am going through and running each of these through the function model-ad-bayes() in model-ad-angio-bayes.R "by hand", but i would like to figure out what is causing this behavior. if you could try one of these out, that would be very helpful. I am at a loss -- every other dataset appears to be behaving just fine.

mwpennell commented 10 years ago

at least some of these clades also gave me a hard time when i ran the ML (but can't find where i wrote those down...)

richfitz commented 10 years ago

What's the actual problem? Error messages, failed convergence?

mwpennell commented 10 years ago

the only real problem was in the very small datasets. I believe lm was choking. All fixed now