apathy

39 ms·

391.
by apathy 10y ago
Re: most data scientists wouldn't change their home page based on 516 data points: Out of curiosity, if you find that in 240 out of 261 small round blue cell tumors, your algorithm can make a better treatment decision than the patholog
392.
by apathy 10y ago
I have a t450s running Ubuntu and it's awesome. Plan on buying a 16GB DIMM, obviously. I wish they'd add more slots for RAM. Also, get the option where you can cram 3 SSDs into it. Then you can swap to "slow RAM" if yo
393.
by apathy 10y ago
Well, yeah. That's why a team of 6-8 people could patrol the sites for 8 hours a day and by morning, there would be plenty of new crap to start all over again. Remember, they were only bothering to whack copyright infringing porn, or
394.
by apathy 10y ago
That's really cool to hear! The customer service people at Xoom were great ones, many of the more colorful characters I've ever met. Ed went off to start a security firm, Bryan moved to LA to star in gay porn, and I dated a crazy
395.
by apathy 10y ago
Wow, that's a terrific article. I don't think anyone at Xoom was in a position to make or break a revolution (maybe on Xoom International, but we'd delegated the whacking duties to the Italians on those sites, if memory serve
396.
by apathy 10y ago
http://sci-hub.bz slash 10.1038/nature18626 may help here. Oh god, a chord plot. There's a harbinger of bogosity if ever there was one. Still... Interferon gamma is a pretty interesting part of the IFN cascade. Immune
397.
by apathy 10y ago
Xoom.com was once a site where, in exchange for free web space (kind of like Geocities), you gave the company permission to spam you. "We Know You'll Love This!" was pretty much our standard line, but we were profitable, in
398.
by apathy 10y ago
Try http://www.r-bayesian-networks.org
399.
by apathy 10y ago
Mapping an alignment gets farmed out to C via rsubread, samtools, etc. as far as annotation, anyone who doesn't use GRanges (or vcfanno, bed tools, etc) is insane. (You shouldn't be aligning locally anyways, it's a waste of
400.
by apathy 10y ago
Seriously? The topics in the second half of the class are what turned R into something notable. People would still be jerking around with Matlab or SAS (and getting poorer results in many cases) if Gentleman hadn't decided that micro
401.
by apathy 10y ago
You should do whatever you like, but remember that domain knowledge and math skills are not mutually exclusive. If the problem you need to solve for a major customer or project happens to be in that 30%, it may come in handy. Linear algebr
402.
by apathy 10y ago
I disagree that RMS is necessarily better than ESL in some meaningful way. The two are complementary. It's like saying a gravel truck is better than a motorcycle: it all depends on what you want to do with it. Re: "do I really nee
403.
by apathy 10y ago
You ran into IPC process limits. The usual way around this is just to run the tasks across a bunch of nodes, using mclapply or openMP (if running C library calls) on each. R is not beautiful but can be coerced into getting things done if ne
404.
by apathy 10y ago
Before 2007, you mean? Because that's when SICP was deprecated for MIT 6.001. Happily, it's still available: http://web.mit.edu/alexmv/6.037/
405.
by apathy 10y ago
Yes and no. Multicore/multisession is not the only option for parallelism in R. The default approach is to use fork() on *nix and... hell, I don't use Windows, so I don't really care how it works there. But there are other
406.
by apathy 10y ago
Think Bayesian vs frequentist. Obviously from the perspective of "optimize against a loss function" they're both special cases, but one is mathematically and philosophically parsimonious, the other... Less so.
407.
by apathy 10y ago
See also https://mitpress.mit.edu/sicp/full-text/sicp/book/node70.htm... for fun background reading. Or if you have an actual CS degree, review from freshman year.
408.
by apathy 10y ago
File issues with Henrik. He actually reads them.
409.
by apathy 10y ago
You may be waiting for a while. The thing that R has going for it is documentation. The thing it has going against it is the monstrosity of the language implementation. Julia is nice if you're coming from Matlab. And it sure as hell is
410.
by apathy 10y ago
Possibly, but you'll have to work, hard, to keep up.
411.
by apathy 10y ago
I disagree re: ESL and RMS, but Harrell's book is superb. It's really more aimed at biostatisticians and clinicians, though. If the math in ESL gives you trouble, you might prefer http://www-bcf.usc.edu/~gareth&#x2
412.
by apathy 10y ago
Or, overfitting strikes again. Except that here the overfit is buried in a bunch of layers. Still -- eventually, especially by reversing the models to generate instead of discriminate, it is possible to be more transparent about what's
413.
by apathy 10y ago
Another excellent point. R, like UNIX, is user friendly. It's just very demanding towards its friends.
414.
by apathy 10y ago
Again, it's sort of like COBOL: it's dying, but until it's dead, there's a lot of money to be made maintaining old macro libraries etc for dinosaur companies. If anyone brings up the FDA... Just don't do that ok?
415.
by apathy 10y ago
This is the best answer I've seen, I wish I'd seen it before I wrote mine. Great explanations. R/S was really meant as "glue" for FORTRAN and C linear algebra libraries (ask Chambers if you don't believe me, an
416.
by apathy 10y ago
All of the above can be done relatively easily by any individual. And like I said, the Revolution guys aren't all bad. It's really just revo R that I find uncompelling. Even that wasn't always true, but Hadley and Rstudio hav
417.
by apathy 10y ago
Yeah, but on the other hand the resulting project has happily eaten research statistics. It's not all bad.
418.
by apathy 10y ago
You just described all fields worth doing research in. If we knew the right answers we wouldn't waste our time implementing the wrong ones. It's not a bad thing: it means there is room for you to discover something genuinely new a
419.
by apathy 10y ago
Do what the author did: take Koller's course and follow the research that interests you. In this case, the propagation through factor graphs is also a good analogy for when you start looking at back propagation in NNs and the autoencod
420.
by apathy 10y ago
I think he was under the impression that using the GPL (instead of a BSD license) would avoid commercial forks. Personally I don't care, Gentleman doesn't care, and the Revolution people (some at least) are pretty good guys. But

More ›