stdbrouw

61 ms·

601.
by stdbrouw 11y ago
Sure, simulate it if you don't believe me. x <- rnorm(1000, mean=50, sd=10) quantile(x, 0.99) mean(sapply(1:20, function(i) { i <- i * 50 quantile(x[i-49:i], 0.99) }))
602.
by stdbrouw 11y ago
Even P99 estimates averaged from many tiny samples are at most inaccurate, they won't be biased. The problem you talk about is one of granularity: you don't want to know how bad things are in general, you want to know how bad thin
603.
by stdbrouw 11y ago
> One is that A/B testing can only lead to a local maxima in the best case scenario. What you mean is: it can only lead to a local maximum in the worst case scenario. Even blind guesses can occasionally lead to a global maximum, and
604.
by stdbrouw 11y ago
The article is useful because it outlines many different ways in which to monitor the performance of a system, many of which are better than just looking at the mean and P99. However, the main thesis that "an average of a percentile is
605.
by stdbrouw 11y ago
Don't confuse yourself with the average user. Facebook and Twitter go where their tests and metrics tell them to go. If an algorithm that highlights important content and hides stuff you probably don't care about makes the average
606.
by stdbrouw 11y ago
> I find the pacing and "bubbliness" exhausting, and the tone bordering on patronizing. The description "intellectual used-car-salesman voice" mentioned in the article is spot on.
607.
by stdbrouw 11y ago
Nitpick: "If you delete the application from your phone, your data is gone for good" is the worst possible way to phrase what is otherwise an excellent feature :-)
608.
by stdbrouw 11y ago
Stacked graphs are really hard to parse. You have all of these lines going up and down even when no upward or downward trend is present. It would probably be better to just create 10 small barcharts (one for each reason) and then put those
609.
by stdbrouw 11y ago
It is certainly possible that you have lost the ability to focus, but why not try this much more straightforward explanation first: you didn't like the last couple of books you've read, which is why had trouble finishing them. It
610.
by stdbrouw 11y ago
Isn't this just how employment works? If you build something and it turns out to have been a total waste of time and energy, you get paid all the same. If you build something and it turns out to save the company $600k a year, you also
611.
by stdbrouw 11y ago
Because it's convenient and can still taste really good.
612.
by stdbrouw 11y ago
Well, a good example would be linear regression. We know pretty much all there is to know about its asymptotic behavior, and can prove that its results are "maximum likelihood estimates." Similarly, Bayesian statistics (a.k.a. pro
613.
by stdbrouw 11y ago
Or akin to PHP's pass-by-reference ampersand, you could have something like `&x + 1` which would expand to `x = x + 1`. Dunno what the impact on code readability/maintainability would be, but it does sound nice. Especially whe
614.
by stdbrouw 11y ago
> I'm not convinced most states should be spending the time or resources setting their own individual educational programs. With the global economy we are in, it's not like the needs of students in Washington state are that dif
615.
by stdbrouw 11y ago
I think you really get into this sort of mess when you want to squeeze the last ounce of predictive performance out of an algorithm (or ensemble of algorithms). When you just want performance that is better than a plain old regression, I&#x
616.
by stdbrouw 11y ago
The author seems to imply that frequentists are not just wrong but borderline malicious – they use all sorts of ad hoc procedures, have no theoretical justification for anything they're doing, and so on and so on. But in fact, there is
617.
by stdbrouw 11y ago
A sequence corresponds to multiple draws from that population, and a statistic (number of successes / number of draws) is a summarization of those draws in a single number. The author isn't really arguing that a binomial distribut
618.
by stdbrouw 11y ago
This is just for sport, right? Otherwise, just rendering a sine wave with Canvas would be a heckuvalot easier.
619.
by stdbrouw 11y ago
Are you serious? Try `ls *.md | xargs -I {} pandoc {} -o {}.pdf`
620.
by stdbrouw 11y ago
Guess we can't keep waiting for the vaporware that is Mou 1.0.
621.
by stdbrouw 11y ago
Some of the shortcuts look better thought out in Caret (in particular the inline file browser), but mostly, it just looks slicker. For 5 bucks, I don't expect it to be "much better", just a little better.
622.
by stdbrouw 11y ago
For me, the point of Markdown (and LaTeX for that matter) is that I can focus on just the writing, not what my document will look like.
623.
by stdbrouw 11y ago
It'd be nice to have it be part of the editor, but `pandoc document.md -o document.pdf` does the trick. http://pandoc.org/getting-started.html
624.
by stdbrouw 11y ago
Friendships form when you repeatedly interact with the same person. This is not likely to happen at a random bar, but it most definitely can happen in a cooking class, when volunteering, playing music with people in a band, routinely gettin
625.
by stdbrouw 11y ago
This just proves my point, though. Gumbo is an HTML parser, not a scraper. The AWS library "will" support EC2 and S3. Again, this is all perfectly normal and it'd be absurd to expect such a young project to have a rich ecosys
626.
by stdbrouw 11y ago
> But to get going with Julia, I just download Julia and go. Until you need to do something very straightforward like web scraping or interacting with AWS and find out that nobody's released a package for that yet, so you're go
627.
by stdbrouw 11y ago
> Python fans are a bit unnerving, they seem to be more vocal among the similar alternatives R, Octave/Matlab, SPSS It'd be difficult to list four languages that are more different than those in your list. There's some ove
628.
by stdbrouw 11y ago
Do you have it set so that it changes colors gradually? It's much easier to get used to if you have it change over the course of an hour instead of instantly.
629.
by stdbrouw 11y ago
I was talking with a social scientist recently who (reluctantly) had to write code in order to run certain statistical analyses. Her solution, to keep code understandable, was to make sure everything could be read as a linear script, in the
630.
by stdbrouw 11y ago
Not all science is theory-based. If biologists discover a new species, I sure hope they're not required to have a theory about why the species "has to exist"; even detailed descriptions and study of its habitat and organs can

More ›