ced

48 ms·

421.
by ced 14y ago
Linux sales are dwindling, sadly. Judging from the "Total Payment" pie charts of past bundles, we went from ~23% (bundle 2) to 12% (bundle 4) to the current ~9%. Total sales across all platforms have remained fairly constant.
422.
by ced 14y ago
The faster-than-light neutrinos probably got explained by a loose cable, sorry. :) Right:) That's exactly what I was referring to: the uncertainty about the model/experimental setup is larger than the N-sigmas uncertainty reported by the
423.
by ced 14y ago
When are Bayesian methods "clearly worse" than frequentist methods, apart from computationally?
424.
by ced 14y ago
Right - MAP may (debatably) be a lousy Bayesian approximation, but it's still Bayesian :) David MacKay's wonderful book made the observation that MAP is a variational method that uses a delta function. "From this perspective, any approximat
425.
by ced 14y ago
Last note: another thing you can do when you have a prior is factor it into your estimator and take the maximum likelihood of the posterior as a point estimate. This is called Maximum a Posteriori (MAP) and called by some people "Bayesian"
426.
by ced 14y ago
3.4 sigmas is the within -model uncertainty - that is, given the assumptions about their instruments and methodologies, the result stands at 3.4 sigma, or about 1/1000. I think that the uncertainty about the model and assumptions is alre
427.
by ced 14y ago
With less than 20 non-overlapping points on the graph, I doubt that I would want to use anything else. Furthermore, if my data has two outliers that are near each other, they may well be indistinguishable in the hexplot from one (or five) c
428.
by ced 14y ago
hopefully it won't look like Japan's I'm curious: what's wrong with Japanese politics?
429.
by ced 14y ago
No. It is possible to ensure that people don't cheat even with FOSS; signing and encryption make it possible to set up a nice relatively-difficult barrier to cheating. How would that work? If I have the source code of Starcraft, I can tur
430.
by ced 14y ago
Related question: wouldn't FOSS ruin competitive online play, by making it impossible for game companies to prevent cheating?
431.
by ced 14y ago
I've looked at your website, and from what I gathered, I would make the same criticism as I made for Anscombe's rule: it's not easy at all to decide what rewards should be, and how to put a price on exploration vs. exploitation. The more I
432.
by ced 14y ago
If you aim to make inferences about which ideas work best, you should pick a sample size prior to the experiment and run the experiment until the sample size is reached. That's not a very Bayesian thing to say. It doesn't matter what samp
433.
by ced 14y ago
Where do I find these small consulting shops?
434.
by ced 14y ago
Computational geometry??? That's a new one for me. Do you mean only linear/convex programming? Incidentally, I would really like to hear about the kind of Real Work that data scientists end up doing with TBs of data, because I'm always fuzz
435.
by ced 14y ago
From Alan Kay's interview: The problem with the Cs, as you probably know if you’ve fooled around in detail with them, is that they’re not quite kosher as far as their arithmetic is concerned. They are supposed to be, but they’re not quite
436.
by ced 14y ago
I'm a long-time Ubuntu user, and I use Wine only when I absolutely have to, because it most definitely does not "just work". Case in point, GP was talking about Photoshop. This is from Wine's page for CS3: First off: don't expect this to w
437.
by ced 14y ago
I like Kent Pitman's explanation of why there cannot be One Universal Equality in Lisp: http://www.nhplace.com/kent/PS/EQUAL.html
438.
by ced 15y ago
I still don't understand how that is a win for consumers . If anything, my limited experience with the stock market tells me that market makers and HFT have been stealing cents on most transactions made by individuals since we've made the
439.
by ced 15y ago
Moreover one of the best things for consumers was when the stock market changed quotes from fractions to cents. Why?
440.
by ced 15y ago
I agree that Rusell and Norvig AI doesn't have much penetration yet. As for Elements of Statistical Learning... That's the canonical textbook for ML. If industry relies on splines, boosting, and support vector machines, then it is really n
441.
by ced 15y ago
my understanding from undergrad and talking with people is that there is a heavy split between academic ML/AI and industrial ML/AI. One uses a lot of raw brainpower, the other uses a lot of base cunning to reduce problems that look like th
442.
by ced 15y ago
Do you have any evidence for that? I don't have any background in pharma, but my background in stats tells me that the odds of passing a controlled study with a placebo would be so low as to be economically infeasible.
443.
by ced 15y ago
How is the hippy currency peer-to-peer? Isn't it just like any other fiat currency?
444.
by ced 15y ago
Thank you for the info. One last question: do you use gender, age, and other customer-specific predictors in your model? The distribution of lambdas for men and women could vary significantly.
445.
by ced 15y ago
OK, that's a good point. Have you considered hierarchical modeling, like in Bayesian Data Analysis? I would have lambda and mu drawn from per-company gamma distributions, and have the parameters of these gammas drawn from global distribut
446.
by ced 15y ago
I was a fan of Gelman-style Bayesian modeling for a while, but it's a lot of work when you don't have a True Model. Have you tried machine learning techniques? You have a ton of predictors available for each customer --- gender, age, price
447.
by ced 15y ago
Furthermore, SBCL type declarations are - Statically checked and removed at compile-time if possible - Checked at runtime otherwise - Not checked at all if you compile with high speed and low safety, yielding close-to-C level of performance
448.
by ced 15y ago
It doesn't have to be irrational. Haskell could easily be better than Common Lisp, but I doubt that the difference is sufficiently large to justify the cost of switching. So I don't.
449.
by ced 15y ago
Does 1-based indexing have any advantage, beyond familiarity to scientists?
450.
by ced 15y ago
Isn't it problematic that 11, 01 and 10 XOR together to get 00? And why not use a simple counter?

More ›