Musings, Explorations, and Announcements

 
 

17 January 2013

At Davis R Users’ Group yesteray, Steve Culman gave us an introduction to the plyr package and how to use it to manipulate data. Here’s his presentation, and the accompanying demonstration script: Steve’s talk is based on this paper by Hadley Wickham in the Journal of Statistical Software. A lot of useful related resources are at Hadley Wickham’s plyr website. We had a quick exchange about using plyr for parallel processing.

Read More…

 
 
 

7 January 2013

This a brief guide to using R in collaborative, social ways. R is a powerful open-source programming language for data analysis, statistics, and visualization, but much of its power derives from a large, engaged community of users. This is an introduction to tools for engaging the community to improve your R code and collaborate with others. (Am I missing anything? Let me know in the comments and I’ll update this guide.

Read More…

 
 
 

7 January 2013

My study system is Sudden Oak Death (SOD) in California Forests, so my examiner [David Rizzo] has given me plenty of readings on the subject. There’s too much for a pithy summary here, but some highlights about the disease’s biology: SOD is caused by Phytophthora ramorum, a water mold in the same family as brown algae. It’s origin is unknown, but 3 lineages are known, two dominant in California and one in Europe.

Read More…

 
 
 

23 December 2012

My examiner on theoretical ecology, Marissa Baskett, suggested that I return to some basic literature on the why of mathematical modeling in ecology. Here are notes on three papers: Summary Modeling purpose must be clearly defined, as this determines whether to focus on generality, realisms, and/or precision Complexity is as often obfuscating as illuminating When prediction is a desired goal, separating calibration from validation is essential Must test robustness of models against different mathematical assumptions, structures, and parameters Many of these assumptions relate to the need to model at a simpler level than reality Disclosure, clarity, and reproducibility in publication are required to justify model use for prediction and policy Model precision is irrelevant if the model fails to address a question of relevance.

Read More…

 
 
 

21 December 2012

Here’s the first of my quals reading summaries, this landscape forest pathology. David Rizzo will be examining me on this topic. Summary Much of our knowledge of pathology is at the organismal level (host-pathogen interactions), but actual management occurs at stand, forest, and regional levels. To understand and manage forest diseases at this scale, we need appropriate epidemiological and landscape ecology tools and frameworks. Gilbert and Hubbell (1996) discusses the differences between traditional, agriculturally-focused plant pathology and pathology at the forest community level, with conservation management goals.

Read More…