Quantum Forest

After lunch of March 3^rd, I posted my last message and made my last visit to an environmentalist forum. It was an interesting experience to present people alternative views, breaking the echo chamber effect. However, it required quite a bit of time that I could not longer justify to myself. Rather than talking about protecting the environment I prefer to spend that time working on it. I am now in the middle of an interesting statistical analysis of birds populations in Tasmanian forests.

After working on the typical descriptive statistics, I tried a log-linear model in Splus, which captured a small part of the variation. I then wanted to try a generalised linear mixed model (like glmm in Genstat), to include random terms in the model. However, the current professional version of Splus does not come with procedures to implement this. Fortunately, I found in Insightful’s site the S+CorrelatedData experimental library that implements glmm. I will give it a try during next week, including first random site effects and then random regressions for sites.

As always, working in a new problem is an excellent learning experience, and a good opportunity to add new statistical tools to my repertoire.