Authors: Brianna and Kelsey
What are we up to lately at Fossilosophy? Good question. Here’s what was on our minds last week.
Guys, there are not very many things as satisfying as going on a paper-downloading spree for something related to your research. There is something to be said for deciding what you need to know, flailing around up to your eyeballs in the literature for awhile, and pulling something coherent out of the mess. Lately Brianna has done this for a few different areas, including the painfully general “(evolutionary) locomotor biomechanics and skeleton stuff!” and a much more specific hunt for statistical quirks in using discriminant analysis to classify fossils. Now, to read and/or skim those ~70 papers…
Kelsey, on the other hand, is probably printing about ten papers per week, and greatly enjoying that every single one of them is in color.
Data Analysis in R
Whenever it comes time to do some data analysis in R, you generally have three situations: 1) You have no idea at all where to begin. Time to start asking books, friends, and the internet, 2) You know all the things you want to do but don’t have all the proper tools, or 3) You know what you want to do and are perfectly capable of writing efficient, elegant code to do so.
If you answered situation 3, that’s very nice. (We haven’t run into that very often yet.) Situation1 happens more often than we would like to admit, but right now we have been thinking about situation 2. You can take one of two general approaches: using the tools you have to get the job done, and teaching yourself new tools. The first choice means doing many things stuff manually, repeating lots of code, resorting to programs like JMP for bits and pieces because it’s so much faster, and just doing things in whatever way you can to get the job done. Sometimes this is what you need, especially if you’re crunched for time because of a deadline. It’s the penny-wise and pound-foolish approach, if you will.
Lately the ladies of Fossilosophy have been aiming for the seond approach, where you go teach yourself the tools you need to do things the right way. It takes way longer and it’s frustrating, but it also builds character (and coding skills). Three cheers for doing things the hard, but proper, way.
Kelsey likes to think of coding as a set of nested dolls or a machine where every part has to be hand-made. Every “gear” is tested as you go, which generally cuts down on debugging later on. The overall operation (say, the regression part of a regression analysis) may be the very last bit of surrounding code you add, after all the parts are moving.
Also, we are fond of giving objects amusing names. This will help you remember all the variables and make readers of your shiny published code smile.
If anyone is interested in a very friendly, straightforward introduction to R book, we liked Getting Started with R: An introduction for biologists, by Andrew Beckerman and Owen Petchey. It is written with great clarity, has good examples to work through, and has just enough tongue-in-cheek humor to make the reading fun.
Bayesian vs. frequentist approaches
It’s a giant can of worms. Make that wormholes…a can of wormholes that, once opened, can send you huddling up under the table faster than trying to wrap your head around what exactly a genus or species is. Though Brianna has enough statistical background to understand the broad ideas behind what’s going on in the Bayesian/frequentist/pragmatist arguments, the whole debate is a little overwhelming and difficult to wrap her head around. (Latest round of mental crisis sparked by an older Oikos post on the matter. Good links and comment section there, too.)
More on this later, after we straighten out our thoughts a little more.
Other fleeting things occupying our attention: how awesome flow charts are, how difficult it is to estimate how long some academic/research task will take to complete, and how great it is being able to bounce ideas off of fellow grad students.