Diaconis and Efron on the bootstrap

An oldie but goodie: Diaconis and Efron, 1983, “Computer Intensive Methods in Statistics” on the bootstrap and other advances in theoretical statistics made possible by the computer. I was particularly struck by a claim that a computation of a bootstrapped confidence interval for a correlation coefficient took less than a second and cost less than a dollar. That seems like a fantastic deal if the alternative is hand computation…

(also, this was written for Scientific American! They’ve gone downhill…)

Half a month’s rain

I’m in London, where according to the Guardian we’re supposed to get half a month’s rain today.

Sounds scary, doesn’t it?  That makes it sound like it’s going to rain fifteen times what it does on a typical rainy day; of course it means that we’re supposed to get fifteen times what London gets on a typical day, including the days when it doesn’t rain.

And in fact the forecast calls for only an inch. (Is that a lot for a single event here? I don’t know – my usual sources for weather data are US-centric.)   Certainly a lot, but not apocalyptic.

And since rain usually comes in storms, it stands to reason that the single biggest rain event in any given month would be a large portion of the rain for the whole month. I’d wager that in a typical month, the biggest rain event is at least a quarter of the total rain for the month.

(Incidentally, there are bookmakers everywhere here. Can I actually bet on the weather?)

Weekly links for May 13

Premise is mapping the produce manifold.

Airbnb is figuring out where you want to go in places you’ve never been.

Diffuse Prior figured out when The Simpsons jumped the shark.

Jim Holt reviewed Mandelbrot’s memoir for the New York Review of Books.

From Smithsonian magazine, Life in the city is essentially one giant math problem.

Visualizations of planar choreographies (Certain symmetric solutions to the n-body problem.) Via hacker news; here’s the paper by James Montaldi and Katrina Steckles.

From Laura McLay (Punk Rock OR), Braess’ paradox in physical systems and in basketball.

Sally Thomason on ultraconserved words at Language Log.

a MathOverflow list of modern mathematical achievements accessible to undergraduates.

Are hot hands in sports real after all?

How often does it happen that theoldest person alive dies?, from math.SE via Hacker News

The paradox of the proof, on Mochizuki’s proof of the ABC conjecture.

Sean Gourley (of quid) on data scientists as the new cartographers.

Name length and job success

The shorter your first name, the bigger the paycheck, says The Ladders, a career website. (via Quartz. This is being publicized as “each extra letter in your name costs you $3,600”. So could I get $10,800 more by going by “Mike”? Seems unlikely – and they are claiming that the effect holds up even with nicknames.

There does seem to be some sort of pattern – average salary peaks for five-letter names, dropping off both for longer and shorter names. Not surprisingly, five letters is pretty close to average name length. I took a look at the Social Security Administration’s list of 1000 most popular names for each sex, from 1960. (The original post was looking at “C-level” employees, who are going to skew somewhat old relative to the rest of the labor force.) Average male name length in this sample was 5.59 letters; average female name length was 5.73 letters. Furthermore, among male names, the shorter names tend to be more common than the longer ones. (This doesn’t hold true for female names.)

My theory, then — which I don’t have the data to test — is that people with more common names tend to do better at the C-level. Perhaps parents who give their kids common names tend to be more conformist and raise their kids with the sort of values that will get them such a job. Or perhaps kids with more common names end up more self-confident, since they’re not constantly thinking that their names are weird — and self-confidence is important for career advancement.

(None of this should be taken to be comments on my own name, Michael, which was very common in the early eighties.)

Weekly links for May 7

Sears Merritt on safe leads in basketball, expanding on Bill James’ method for telling when a lead is safe.

Kevin Jamieson makes a two-dimensional map of styles of beer. I’d like to see this for wine.

You can buy a 3D-printed triple gear (three gears which touch in pairs and yet can all turn simultaneously! or
read about how Saul Schleimer and Henry Segerman figured out how to make one.

Ben Orlin, a math teacher, writes about what it feels like to be bad at math.

chartsnthings made some charts of skill and chance in the NFL draft.

Melinda Theilbar explained how folklore can be dangerous to one’s business.

Ravi Vakil wrote about the mathematics of (a certain kind of) doodling.

Weekly links for April 29 (one day late)

From Josh Wills at Cloudera, a post on reservoir sampling.

Evelyn Lamb has compiled a list of mathy ladies to follow on Twitter.

Stephen Wolfram (and presumably part of his army of people working for him) have some interesting visualizations of Data Science of the Facebook world.

Brian Hayes maps the Hilbert curve.

Dana Mackenzie at Slate writes on the mathematics of jury sizes. Also at Slate, Phil Plait writes for the Bad Astronomy blog on the analemma.

How to sort comments intelligently and this post on Bayesian methods for multi-armed bandits are part of Cam Davidson-Pilon’s book Probabilistic Programming and Bayesian Methods for Hackers. I found Davidson-Pilon via his list of machine learning counterexamples.

Kenneth Appel (of four colors suffice fame) died.