Stack Exchange
log in

Matthew Drury

 Discussion on answer by Flater: Multi…

Imported from a comment discussion on softwareengineering.stac...
Matthew Drury
Jun 29, 2023 13:56
As a real life example of the RaiseToPower concept, in the Redshift SQL dialect (maybe others as well), datediff(unit, b, a) does not map conceptually to b - a, violating centuries of arithmatic convention, but maybe satisfying someone who wishes subtraction followed the ordering of the number line?
 

 Ten fold

CrossValidated's general room for gossip, grumbles, and idle c...
Matthew Drury
Jul 20, 2018 18:42
I've got some chores to do, but will try to work on an answer this weekend.
Matthew Drury
Jul 20, 2018 18:41
I just saw that!
Matthew Drury
Jul 12, 2018 21:33
And thanks for the link to the meta thread, I somehow didn't catch that one yet.
Matthew Drury
Jul 12, 2018 21:33
Oh yes, that's good. Your answer there is more or less what I am spewing repeatedly. I'll start including a link to it.
Matthew Drury
Jul 12, 2018 20:47
I fell like I, at least, am becoming a self parody on this point.
Matthew Drury
Jul 12, 2018 20:46
Do / should we have a canonical answer about thresholding in binary classifiaction.
Matthew Drury
Jul 6, 2018 21:40
@Sycorax Oh, I'll be going outside. I've already stacked a backlog of skateboard decks and mapped out all the spots in Seattle. Ready to go.
Matthew Drury
Jul 6, 2018 18:52
So if anyone has some thoughts on how I could best help out while I'm on break, im open to suggestions.
Matthew Drury
Jul 6, 2018 18:51
I'd also happily use some of that time in service to the site. I've been guilty of comments as answers for some time, since most of my available time for the site has been on the bus lately.
Matthew Drury
Jul 6, 2018 18:51
@Sycorax Beginning the end of next week, I'm on break from work for two months. I'd be happy to put together a canonical answer on back propagation, but not until then!
Matthew Drury
Jun 22, 2018 14:46
There's also a really strong python community in Seattle (PuPPY) which is very welcoming and inclusive. There's a healthy stats / data science / machine learning subgroup of that community (ATOM) that I help organize. This has been a great source of friends for me.
Matthew Drury
Jun 22, 2018 14:45
@MaskedMan Ok. Thanks for the assist.
Matthew Drury
Jun 22, 2018 14:45
@Sycorax

I like it in seattle. I've lived in a bunch of different places, and it's probably my favorite or second favorite place I've lived. It's a big city, there are lots of busses so you can get away without driving, there are TONS of parks, and outside of the city there are mountains and nature. The pacific northwest is just a wonderful climate.

There are lots of jobs for the taking. The obvious ones are at amazon, microsoft, facebook, these companies are always hiring programmers and data scientists (and all the better if you are both). There are also many smaller companies, sta…
Matthew Drury
Jun 21, 2018 14:36
Well done.
Matthew Drury
Jun 21, 2018 14:36
@Sycorax I shared your NN thread on the Seattle ML user group. You're getting some serious virtual hi-fives.
Matthew Drury
Oct 24, 2017 00:21
Ahh, got it, thanks fot the link. I can offer you an answer later.
Matthew Drury
Oct 24, 2017 00:11
@kmace It's largely theoretical, a conceptual tool. That would make a good question for the main site, where you would probably get a thoughtful answer.
Matthew Drury
Oct 3, 2017 21:59
Unfortunately, R using a hash table is giving R too much credit. :(
Matthew Drury
Oct 3, 2017 21:59
Looks like table calls tabulate but precomputes the number of bins. So the call to table is looking through the data at least twice.
Matthew Drury
Oct 3, 2017 21:58
I took a quick look at the source.
Matthew Drury
Oct 3, 2017 21:52
@whuber Interesting. On the other hand, if table is building such a hash, wouldn't it be something like number -> count. In which case there is only six possible keys, so once all are seen at least once, the data structure would not have to be resized?
Matthew Drury
Aug 28, 2017 16:59
Man, I've been hanging out here for two years and I just learned that new users cannot upvote!
Matthew Drury
Aug 28, 2017 04:23
It's like saying: I accept that this is the answer, but I am saddened by it.
Matthew Drury
Aug 28, 2017 04:23
It's so weird when someone accepts your answer and does not upvote it.
Matthew Drury
Aug 23, 2017 04:39
Not that we would.
Matthew Drury
Aug 23, 2017 04:39
If we judge them harshly, it doesnt matter much.
Matthew Drury
Aug 23, 2017 04:39
We're probably playing the role of the anonymous collective.
Matthew Drury
Aug 23, 2017 04:38
Sometimes people are very apprehensive about exposing vulnerability to people with authority over them.
Matthew Drury
Aug 23, 2017 01:56
It seems like you guys understood what happened a little better than I, but I'd be happy to raise it if that's appropriate.
Matthew Drury
Aug 23, 2017 01:55
It's not super important to me, but if I've unearthed an edge case that the dev's didn't intend it's probably worth raising it.
Matthew Drury
Aug 23, 2017 01:54
Thanks everyone.
Matthew Drury
Aug 21, 2017 22:13
I've never hit the reputation cap before!
Matthew Drury
Aug 17, 2017 02:04
Maybe not that many people read meta, but the people that do are likely to be those that would like to help out with duplicate questions and other housekeeping tasks.
Matthew Drury
Aug 17, 2017 02:03
I would really appreciate that meta thread.
Matthew Drury
Aug 15, 2017 13:37
I'm hoping to get better at doing things around the site that are not just answering questions. It's silly, but the search tips were non obvious to me.
Matthew Drury
Aug 15, 2017 13:36
Thanks guys, that's very helpful.
Matthew Drury
Aug 15, 2017 02:24
Thoughts?
Matthew Drury
Aug 15, 2017 02:24
So it would serve as a simple reference to consult when in that position of: I want to vote to close this as a duplicate, but what question...
Matthew Drury
Aug 15, 2017 02:23
Where we can contribute and discuss canonical duplicates.
Matthew Drury
Aug 15, 2017 02:23
Like the top ten reasons to close
Matthew Drury
Aug 15, 2017 02:23
I'm thinking of starting a meta question: List of canonical answers to common questions, or whatnot.
Matthew Drury
Aug 15, 2017 02:23
Amoeba suggested some time ago having canonical duplicates, and I think there are a fair number of questions that would benefit from having a canonical representative.
Matthew Drury
Aug 15, 2017 02:22
I often come across questions that I strongly suspect are duplicates, but don't have any specific duplicate in mind.
Matthew Drury
Feb 28, 2017 18:52
We're having lots of pain at work as well.
Matthew Drury
Feb 28, 2017 18:51
A heads up that AWS is not working today, so the images on the site not displaying seem to be related to that.
Matthew Drury
Dec 10, 2016 16:44
The story seems fuzzier than I would have guessed.
Matthew Drury
Dec 10, 2016 16:44
Hi everyone. Thanks you for all your contributions to my regularization question!
 

 Discussion on answer by daniero: Writ…

Imported from a comment discussion on codegolf.stackexchange.c...
Matthew Drury
Jan 30, 2018 02:26
I teach a Machine Learning class, and I just used this answer to make multiple points (you need a baseline for comparison, classification accuracy can be a pretty meaningless metric). It's a good answer with an important point.
 

 Discussion on question by Justin Thym…

Imported from a comment discussion on stats.stackexchange.com/...
Matthew Drury
Aug 29, 2017 10:04
It might be more democratic to say what field we do think addresses your thoughts. The word "after" is a temporal concept, and afaik is not part if statistics. Your question may be better posed to physisists of philosophers, even with its statistical content.