Hi class,
Here is an example script following Zuur’s Chapter 1 from his new book ( Chap1_jtf.r ). I added a few things, including using the coda package to plot some of the BUGS output. The data on feeding of oyster catchers is in the file OystercatcherData.txt (from Zuur’s website http://highstat.com . Also, here is an Appendix from “Bayesian Modeling Using WinBUGS,” by Ioannis Ntzoufras, on how to use the R package “coda”.
Enjoy your weekend!
Jack
Hi class,
The author of the lme4 package has an online book that is worth taking a look at. The latest and greatest version resides in the R-forge web site, and you should load lme4 from there! Some code and the chapters he has put up on the R-forge site are in the zip document Bates_MixedEffectsModels.zip . We’ll talk about the book and some new commands on Tuesday.
Jack
Hi class,
Here is an interesting example from Ben Bolker using the owls dataset. He shows how to use an observation level group effect (to effectively change to a Poisson-lognormal distribution), how to predict the expected values of a lme4 model, and an interesting way to plot expected vs. observed with confidence intervals. I might have to learn ggplot2! Enjoy!
Here are the pdf and the script I did in class:
Jack
Hi class,
I worked on some more models today, and found one that is better. All issues are not resolved, but some are.
New script is now Chapter13_jtf_v2.r
Jack
Hi class,
Here is my modified script for Chapter 13 – Chapter13_jtf.r
Can’t quite get to everything we might want to do. Here is a paper that talks about GLMMs: Bolker et al. 2009 . It goes into some detail about how each pacakge (or program) solves the problem. Apparently, there is real difficulty in solving GLMMs and GAMMs. The difficulty seems to be that the solution to the fixed effects have to be integrated over the random effects. This is not exact, but there are various approximations. PQL (penalized quasi-likelihood) is worst, LaPlace Approximation is better, etc.
Also, look at this link (http://glmm.wikidot.com/pkg-comparison#GLMMs), which lists the various commands and what you can do with them. For example, it looks like glmmPQL is the only command that can do correlation structures.
Have fun!
Jack
Hi class,
Here is the final script for Chapter 12 (Chapter12_jtf_v2.r). I made a few changes in class and after class, I figured out why I was confused about the QIC and yags. It looks like yags (using QIC) and geeglm (using anova) come to different conclusions about whether to drop the interaction. The QIC says don’t, and the anova says do. It’s hard to know if the problem is the software (yags vs geeglm) or the metric (likelihood ratio vs. QIC).
I think the bottom line on GEE is that it is very useful to try, especially if you can’t get just the right distribution to work. It can mimic mixed effects (using intraclass correlation), can account for various other correlation structures (like AR1) and may be able to give you most of what you want. I’d stick to the anova approach, unless you want to look at the references in detail.
Jack
Hi class,
The new scripts are up:
Also, in Chapter 8, I forgot to talk about the likelihood functions for each distribution. That is what R uses to figure out the likelihood of each observation given the parameters.
Jack
Hi class,
I went over Chapter 7 again, and made a few changes. The new script is Chapter7_jtf_v2.r
Jack
Hi class,
The script for Chapter 7 is now online ( Chapter7_jtf.r ). Lots of stuff going on in this chapter. Enjoy!
Jack
Hi All,
Here is the script ( Chapter6_jtf.new.r ) I ended up with at the end of class today (I was trying to plot a smoother through the residuals of the M3.gls model, but it didn’t actually work until after class) . I have also uploaded the slides I did to explain ARMA models. Tomorrow, we’ll work with some of your data and try to diagnose what to do with the random part of the model (fix heterogeneous variances, add random effects, add autocorrelation, or something else we haven’t gotten to yet).
Jack