Eric’s Enlightenment for Wednesday, May 20, 2015
May 20, 2015 Leave a comment
- A common but bad criticism of basketball analytics is that statistics cannot capture the effect of teamwork when assessing the value of a player. Dan Rosenbaum wrote a great article on how adjusted plus/minus accomplishes this goal.
- Citing Dan’s work above, Neil Paine used adjusted plus/minus (APM) to show why Jason Collins was one of the top defensive centres in the NBA and the most underrated player of the last 15 years of his career. When Neil mentions regularized APM (RAPM) in the third-to-last paragraph, he calls it a Bayesian version of APM. Most statisticians are more familiar with the term ridge regression, which is one type of regression that penalizes the inclusion of too many redundant predictors. Make sure to check out that great plot of actual RAPM vs. expected PER at the bottom of the article.
- In a 33-page article that was published on 2015-05-14 in Physical Review Letters, only the first 9 pages describes the research done for the article; the other 24 pages were used to list its 5,514 authors – setting a record for the largest known number of authors for a single research article. Hyperauthorship is common in physics, but not – apparently – in biology. (Hat Tip: Tyler Cowen)
- Brandon Findlay explains why methanol/water mixtures make great cooling baths. He wrote a very thorough follow-up blog post on how to make them, and he includes photos to aid the demonstration.