Archive for the ‘Uncategorized’ Category

The truth is stranger than fiction

27th September, 2010 Leave a comment

It turns out that when you search Google for the phrase 4D pie chart, this blog is only the second link.  The number one spot is taken by a tutorial on how to create animated pie charts with Maxon Computer’s Cinema 4D package.  I can’t detect any hint of sarcasm in the instructions.  Hang your head in shame, Maxon Computer employee.

Tags: ,

Dotplot on xkcd

27th September, 2010 Leave a comment

There’s a nice example of a dotplot on xkcd today.  Hand drawn lines aside, it’s well drawn.  The only obvious thing wrong with it is the scale.  If you’re going to log-transform your data, then the axes labels should show the untransformed amounts.  So the numbers -17 to -5 along the top x-axis should really be e^-17 to e^-5, or even better, number that are based 10, because they are much easier to understand.  For example, compare:

“Stochastic is used e to the power 14 times as often as f*cking stochastic.”

“Stochastic is used 10 to the power 6 times as often as f*cking stochastic.”


Variability of biomarkers in volunteer studies

9th September, 2010 Leave a comment

Variability of biomarkers in volunteer studies: The biological component, of which I am co-author, is now published in Toxicology letters.  It’s a simple, straightforward look at calculating variability in half-lives of chemicals.

Tags: ,

Pie charts over at Juice

3rd September, 2010 Leave a comment

The Juice Analytics blog (incidentally, one of the few corporate blogs worth reading) has a makeover of the Federal IT dashboard, including a discussion of 3D pie charts.  Only 3 dimensions?  Bah!


World fame awaits

28th August, 2010 Leave a comment

After accepting the paper last November, Occupational and Environmental Medicine have finally published the paper that I co-authored: Trends in blood lead levels in UK workers, 1995 to 2007.  The gist of it is that lead is nasty, toxic stuff and we wanted to know how much of it workers get exposed to.  Alongside lead, we looked at haemoglobin levels (which supposedly decreases with increased lead levels) and zinc protoporphyrin (ZPP for short, which supposedly increases in cases of lead poisoning).  The main analysis is a linear mixed effects model, which is a variation on a bog standard regression, letting us quantify how much variation in lead levels was due to differences for a given person, and how much was due to differences between people.  My favourite bit of the paper is Figure 2B.  Given a squint and a dirty mind, it’s full-on Rorschach rude.

zpp vs lead, penis-based correlation