Wednesday, July 1, 2015

Wisconsin Butterflies - 1st Week July

Once again, bold spp. are in Dane Co. This is a great time to look for butterflies!
  • gray copper
  • Edward's hairstreak
  • banded hairstreak
  • hickory hairstreak
  • juniper hairstreak
  • swamp metalmark
  • American snout
  • meadow fritillary
  • gray comma
  • tawny emperor
  • common wood-nymph
  • silver-spotted skipper
  • Horace's duskywing
  • funereal duskywing
  • wild indigo duskywing
  • common sootywing
  • little glassywing
  • Delaware skipper
  • Byssus skipper
  • mulberry wing
  • dion skipper
  • Brazilian skipper

Monday, June 22, 2015

Options for 3D Graphing in R

I've now come to a point in my graph-making where I'm delving into the realm of 3-D visualization. In this specific case, I'm looking at the results of modeling an interaction of 2 predictor variables.

Another specific for me: this is with my hierarchical Bayes fitting with INLA, so in many examples online I find easier ways to predict values...for this case, not so much. For example, in many cases there are cool ways to project your model's predictions onto a grid, but if I need to impute values for hypothetical points, I need to do that on my model's run. So, the best I can work with for inputs into any graphical output for my particular data is an already-generated data.frame or matrix from my model with x,y and z's. The INLA model object returns fitted values, so I use those for "real" data points. If I want to input hypothetical predictor values, I set my response variable for those values to NA and let the program fill in the modeled value.

I'll start with my favorite: scatter3d from Rcmdr. (A note related to the package - for some reason, in Rstudio/one of the newer versions of R, this doesn't show up in the list if you try to search for it. Yet, if you type it anyway and try to install, it looks like it can be found. It doesn't seem to install all its dependencies on the first run, though. It also is a GUI so it will launch its own window, but you can do everything in R with the library() call.) Here, I take a subset of the modeled values and plot them...very easily! I just feed the function my x, y and z's and not only does it plot a point cloud, it gives me a surface too. I can specify what kind of surface I want as well (linear, quadratic, smooth, additive...)

The rest...
  • scatterplot3d() with plane3d - as far as I can tell, the plane can only be based on a glm, so it's always linear or "flat."

Sunday, June 21, 2015

Butterflies - Last Week June

As before, bold species are found in Dane Co. (i.e. most of them!)
  • bronze copper
  • Acadian hairstreak
  • silvery checkerspot
  • northern crescent
  • Baltimore checkerspot
  • Compton tortoiseshell
  • white admiral
  • hackberry emperor
  • northern pearly-eye
  • eyed brown
  • smoky-eyed brown
  • Poweshiek skipperling
  • European skipper
  • Ottoe skipper
  • long dash
  • two-spotted skipper
Happy butterfly watching!

Saturday, June 20, 2015

Lawn Mower's Mushroom

I found these tiny mushrooms, so named because they're so common in ordinary lawns. Look around and you'll probably find them!
Panaeolus or Psathyrella?

Thursday, June 18, 2015


"The" paper about INLA for Bayesian stats starts out talking about the class of structured additive regression models. Fortunately, for non-statisticians like me, they throw me a bone and give me a list of examples from which I can pick out a few that I recognize.
  • (generalized) linear models
  • (generalized) additive models
  • smoothing spline models
  • state space models
  • semiparametric regression
  • spatial/temporal models
  • log-Gaussian Cox processes
  • geostatistical/additive models
The focus is on latent Gaussian models (about which there will be an entire conference this fall; the postdoc I work with and I have joked about going). As the name would imply, the latent field is Gaussian, "controlled by a few hyperparameters and with non-Gaussian response variable" (Rue et al. 2009).

Tuesday, June 16, 2015

Wisconsin Butterflies - 3rd Week June

Species in bold are ones that can be expected in Dane Co., WI where I live.
  • little yellow
  • northern blue
  • greenish blue
  • silver-bordered fritillary
  • Harris' checkerspot
  • tawny crescent
  • red-spotted admiral
  • southern cloudywing

Monday, June 15, 2015

Runtime for Null

(Note to self) null model run time for the whole set is approx 15 min