dxbydt

33 ms·

301.
by dxbydt 7y ago
Hey, thank you so much for posting this! I picked a random problem ( the one on logs) in that paper and solved it under 1 minute. I feel like a million bucks now!
302.
No stable distributions in finance please(arxiv.org)
1 points by dxbydt 7y ago|0 comments
303.
by dxbydt 7y ago
> The most predictive stat for educational attainment is household income and parental educational attainment There is weak correlation with income & a stronger correlation with parental educational attainment. If you club them up &a
304.
by dxbydt 7y ago
>why not just ask them????? Mostly because revealed preferences. https://en.wikipedia.org/wiki/Revealed_preference
305.
by dxbydt 7y ago
The 5 user Nielsen test referred to in this article is quite inaccurate. If ( and that's a very big if) the users are IID, then yes, 5 users is all you need. But your users from Russia aren't the users from USA aren't the use
306.
by dxbydt 7y ago
> If you took away the community entirely, would it be any less useful? so for a while, due to strange quirk of circumstance, i had the inside track of a profitable adult website. it was your standard “streaming adult videos, no ads, cha
307.
by dxbydt 7y ago
Key findings are here - https://bit.ly/2OqMoyD At a 5% significance level, the hazard ratios posted there are clearly not significant. Most of them are reasonably close to 1.0 - so there is basically no difference in mortal
308.
by dxbydt 7y ago
> is ultimately a silly and misguided proposal No its not. Its analogous to quantifying an F test with watts. Nothing wrong with that. I can give a billion row dataset to a dozen people. John assumes normality, samples 100 rows, runs a l
309.
by dxbydt 7y ago
no worries. On your github you have all the point clouds, so I’ll give it a shot one of these days. If you mathematize the distribution, you get a lot more mileage for your results because you get interpretation for free. Changing moments (
310.
by dxbydt 7y ago
> The distribution of points in a point cloud Would you have a canonical name for this distribution ? If you try matching log likelihoods, what parametric family does it resemble ? Briefly, given one of the canonical two dozen (uni/
311.
by dxbydt 7y ago
Its a very well studied dataset in (graduate) Stats. So many contingency tables where you correlate onset of myopia - myopia vs Asian/non-Asian, myopia verus Male/Female, with/without computer, w/wo smartphone, w/wo
312.
by dxbydt 7y ago
Tech for testing is a good idea. You get the same adrenaline rush playing “race the teacher” type math games as in a live classroom test. You are timed, you know exactly how long you work on each problem, you can identify your strengths &am
313.
by dxbydt 7y ago
I did a bunch of work on managing/anticipating the power draw. ( https://news.ycombinator.com/item?id=13747481 ). Something nobody talks about is how grossly wasteful & inefficient Hadoop is. There's a bunch o
314.
Statistics in Ai(reddit.com)
3 points by dxbydt 7y ago|0 comments
315.
by dxbydt 7y ago
It runs both ways, yes. In the forward direction, most countries don't play fair and if they have the chance they will wipe out the US as soon as possible. In the other direction, the US doesn't play fair either and will do the sa
316.
by dxbydt 7y ago
I’m curious why you would think that. afaik, Taleb has repeatedly & explicitly stated “no probability without ergodicity”, which is pretty much a frequentist mantra. The problem with non-statisticians like Nate Silvers (who holds a BS i
317.
by dxbydt 7y ago
> it’s definitely possible to quantify how long it would most probably take to succeed yes. if success means to roll a 1, and anything else is a failure, you will on average roll the die 6 times before you succeed. E(x) = 1/p for a
318.
by dxbydt 7y ago
“future success” does not have to be preceded by past “string of failures”. Like, at all. That’s just classic gambler fallacy. If you want an example of reverse gambler fallacy, see here - https://news.ycombinator.com/reply
319.
Shit Sandwich(statmodeling.stat.columbia.edu)
1 points by dxbydt 7y ago|0 comments
320.
by dxbydt 7y ago
How exactly does a Galton board blow through the CLT ?
321.
by dxbydt 7y ago
> what problems people solve that they need to call IO in the middle of Dijkstra’s algorithm. I worked on several problems of this nature at Twitter in 2012. Hopefully there’s a better way to solve them in 2019...prolly not, but maybe. s
322.
The biggest ways people waste money(wsj.com)
4 points by dxbydt 7y ago|0 comments
323.
by dxbydt 7y ago
>proliferation of mobile devices makes people dumber In the very 1st week of my PhD studies, we had a host of eminent researchers advising us on how to succeed in graduate studies. One of those professors put up a slide that had a giant
324.
by dxbydt 7y ago
Sometimes I wish there was a website where immigrants could list things that are far superior in their country of origin than the one they immigrated to. It’d be very useful for other immigrants so they can temper expectations before commi
325.
Ergodicity(ergodicityeconomics.com)
1 points by dxbydt 7y ago|0 comments
326.
by dxbydt 7y ago
I think you & the parent are both correct, despite holding opinions that are poles apart. Certainly in the kinds of things we are allowed to watch, we are more free than ever before. For instance, wikipedia says the adult entertainment
327.
I’m No Good at Math(qz.com)
2 points by dxbydt 7y ago|0 comments
328.
by dxbydt 7y ago
this is classic “mistaking the map for the territory”. whether the code is in python or matlab matters little. the code is just a supplement to the main material which is still math. trying to understand that math by reading code is activel
329.
by dxbydt 7y ago
it’s a good undergrad book. also on that page he says lights out game is a “linear algebra problem”. It’s actually not. incidentally I worked on Lights out for my thesis, and my advisor made a whole career out of lights out! That game relat
330.
by dxbydt 7y ago
> general inversion of any MxN matrix ??! is there such a thing ? only nonsingular square matrices can be inverted. a general mxn matrix may have a “left inverse” or a “right inverse”, but i don’t think your code is computing that.

More ›