Sunday, February 9, 2014

Empirical Bayes Bakeoff

The ISI has a new journal called Stat,  I'm not sure this is a good marketing appellation in the modern world of search engines, but it has an interesting premise.  It promises quick turnaround, prompt electronic publication and a sanctioned blog forum in which readers can comment on published articles.  It isn't intended for long expository pieces, rather it is meant for pithier, perhaps more provocative pieces that wouldn't fit well in the current journals.  This is a niche that Biometrika's Miscellanea section used to fill.  I decided to give this a whirl recently, submitted a 5 page paper comparing some empirical Bayes methods in several simulation settings of the Johnstone and Silverman "needles in haystacks" type.  In a couple of weeks I received a favorable, and very well informed referee report* and the paper was accepted.  It appeared after a bit of copy-editing a week or two later.  Whether it attracts any Statblog attention remains to be seen,  but I certainly hope so.  It is freely available at:http://onlinelibrary.wiley.com/doi/10.1002/sta4.v3.1/issuetoc  for a while at least.


* In the realm of referee reports, "favorable" and "well informed" are synonyms, so this sentence was a bit redundant.

Saturday, February 8, 2014

Winter Warming?

With all the complaining about the winter weather this year,  I became curious about trends in winter
weather in my home town.  So here is a plot of the last 100 years, or so, of minimum January temperatures in Grand Forks, North Dakota.  The three black lines are regression quartile fits
and the red line is the mean regression fit.  The mean and median fits both indicate that there is roughly a 0.1  degree F per year warming trend over the entire period -- so 6 degrees F since I was in first grade.   This year the minimum was -30F, so quite bit below the median trend line.