Monday December 20, 2010
Ricky Gervais, nice one, in the blog section of the WSJ, http://blogs.wsj.com/speakeasy/2010/12/19/a-holiday-message-from-ricky-gervais-why-im-an-atheist/
Why don’t I believe in God? No, no no, why do YOU believe in God? Surely the burden of proof is on the believer. You started all this. If I came up to you and said, “Why don’t you believe I can fly?” You’d say, “Why would I?” I’d reply, “Because it’s a matter of faith.” If I then said, “Prove I can’t fly. Prove I can’t fly see, see, you can’t prove it can you?” You’d probably either walk away, call security or throw me out of the window and shout, ‘’F—ing fly then you lunatic.”
Sunday December 19, 2010
Aquaplaning up the 405! Rainy Sunday night in LA.
Uneventful flight down from SFO. 757. Seat 2A. 28R departure, no x-rwy ops at SFO due to wind associated with this foul weather. Cloud for most of the flight.
Atypical arrival pattern to Lax, south and then a left turn for final, almost always we fly Malibu thence Santa Monica with a right turn over downtown for final. Pouring rain at touchdown and taxi.
Saturday December 18, 2010
A report on the rebuild of Terminal 2 at SFO, to be used by American and Virgin America.
Only 3 terminals have been designed and constructed in the US since 9/11.
Monday December 6, 2010
Brad Efron continues his long exploration of the implications of James-Stein, empirical Bayes etc: tomorrow (Tuesday) in Stanford Stats, 4.15pm.
Tweedie’s formula and selection bias
Suppose the statistician observes some large number of estimates z_i each of which is unbiased for its unobserved true mean \mu_i. The largest say 10 of the z_i are likely to substantially overestimate their corresponding \mu_i, this being “selection bias,” or “regression to the mean.” Tweedie’s formula, first reported by Robbins in 1956, offers a simple empirical Bayes approach for correcting selection bias. I will discuss its merits and limitations, in terms of some examples and a little bit of theory relating to the James–Stein estimator.
Interesting take on “selection bias”, a little different to the way the term gets used in polisci/econ etc. For instance, one might substitute the words “measurement error” for “selection bias” and I imagine that a lot of the results still apply.
Thursday December 2, 2010
I taught for a week at Juan March in Madrid, which was great fun.
My family came over as well and we spent the week of Thanksgiving back poking around: Barcelona, up into the south of France via the Pyrenees, San Sebastian and Bilbao, then back to Madrid. Just over 2000kms, 7 days on the road. Unforgettable, even in late November.
View Larger Map