rebeccmeister: (bikegirl)
I was recently re-reminded about this cookbook for eating well on a food stamp budget. I should buy a copy to support the project.

I guess there's a new ggplot2 book in the works, which is great news because the previous one is lagging too far behind the current implementation. It's going to include a chapter on programming with ggplot2, which should be fantastically helpful for beginner-intermediate level users looking to improve the efficiency of their workflow. I can't tell you how many different pieces of my recent analyses could use some work in this department, but unfortunately I can't always make the time and space to improve this meta-aspect of my work. On the flipside, I'm still in agreement with myself that Wickham's tools for R users have helped make R incredibly useful for biostatisticians, both in terms of data visualization and data manipulation. Even with my currently cumbersome scripts, I'm able to whip through carefully customized analyses pretty quickly.

This, from McSweeney's, makes me a little tired: Required Reading Essay Questions Written by a First-Year Adjunct Who Does Not Have the Time or Wherewithal to do the Required Reading. Only funny because it's sadly true.

A friend who has been going through insanely hellacious cancer treatments recently posted a link to information about the Fallacy of Relative Privation. I'm still chewing over the idea, but I'm grateful for the introduction
rebeccmeister: (bikegirl)
So, I run the same flavor of somewhat cumbersome statistical analysis in R, over and over again, to the point where it would be really nice to automate certain aspects of the process, especially the aspect where I wind up doing a bunch of annoyingly fiddly formatting in a Word document at the end, after retyping a huge bunch of numbers in a crosseyed fashion (I'm fairly good and anal-retentive about this, but you can see how a person would make annoying mistakes that could be avoided). This eats up hours and hours of my life at times.

At the moment, I'm trying out the 'xtable' package, which will generate output in LaTeX or html. I don't have any LaTeX experience, so I'm going with html for the moment while I try and see if this whole enchilada will work for me. Which means that it looks like I'll be getting to embed a bunch of html tags into sections of my R code, which is just UGLY. Grr.

Plus, character manipulation. I need to compact down my F-statistics, so cells in the table look like so:

Fa,b=c,
P=d


Where the lowercase letters are numbers drawn from specific places, and the bolding of the whole thing is conditional on the P-value (bold if P < 0.05; italicized if 0.05 < P < 0.10)

It's just...ugly.

/whining
rebeccmeister: (bikegirl)
Scork* these days involves a lot of time in front of the computer, managing data. There's part of me that's tempted to start writing R-related entries as I continue to expand my R capabilities, but it also occurs to me that such efforts would probably be most effective in conjunction with some of the other R-related resources I've developed, which are currently offline because of the nuisances associated with .edu-based web hosting and life as an academic. Basically, as a postdoc I'm not at any academic institution for long enough to bother with getting web space at that institution, and one can't exactly obtain an .edu domain independently. Somebody needs to fix this situation, please.

Anyway. I spent most of Friday trying to figure out how to run a statistical analysis in R known as response-surface modeling. cut for lengthy rambling about statistical minutae )
rebeccmeister: (Default)
ASU has this great resource--a statistics help desk. I can take virtually any statistical question to them and they will help me figure it out. Which is great, because I know exactly what statistical test I need to do, but I don't know what software package to use to do it. For any stats geeks out there, it's a nested, unbalanced ANOVA. Zar recommends using a computer because the math is tricky, and I agree. With the whole computer switcharoo, most of my statistical knowledge relies on SAS, and I don't have any notes on nested ANOVAs with SAS. So on Monday, I'm off to the stats helpdesk. Whee!

Oh, I should note--the last time I went there, it was great to watch the stats nerd/expert in action--sort of like watching an experienced hunter stalk his/her prey.

Profile

rebeccmeister: (Default)
rebeccmeister

July 2025

S M T W T F S
  1 2345
6 7 8 910 11 12
13 14 15 16 17 1819
2021 22 23 24 25 26
2728293031  

Syndicate

RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Jul. 27th, 2025 01:19 pm
Powered by Dreamwidth Studios