« first day (1872 days earlier)      last day (3139 days later) » 

12:01 AM
@amoeba It seems to be better known in the UK; it's an older expression, Sir, but it checks out.
@Gabriel Sure looks like a variance to me. Maybe you should comment about it under the post.
@Chris Terry is very good at rendering complicated subjects accessible, but I have to admit, even when he talks about stats and probability he often goes above my head as well. Then again, I've also learned a few useful things from reading what he has to say (or in one case, relearned things I'd forgotten) .
@Erik I definitely agree with gung's point about a post that's already at the top; if there's a time a trivial edit can do no discernable harm it's when the post was already bumped to the top and posts are coming slow enough that you can fit an edit in before more posts arrive. I often take advantage of that to jump in and fix the small niggle or two I'd otherwise leave go.
@ChrisC looking at the first half of the post you pointed to it looks like (more or less) a pretty standard mathematical introduction to probability, which I would definitely benefit from paying attention to ... but probably won't have time for ... at least for some weeks.
 
 
1 hour later…
1:16 AM
@ChrisC Thanks for the pointer to Terry's blog. I had, what i consider, a rather poor experience in my measure theoretic probability class, so I hope to follow along with this.
@Silverfish Is search optimization why you edited the title of this question in the way you did: stats.stackexchange.com/questions/175040/…
 
 
11 hours later…
12:12 PM
@MatthewDrury Partly, yes - if nobody can find a question then they can't be helped by it! (If the question was reopened I might edit to put "caret" in there, as per Nick Cox.)
The problem with "What does this symbol mean?" questions is that if someone else has the exact same problem, but doesn't know what the symbol is called, it's almost impossible to search for!
(I know when I first saw $\hat y$ I didn't think of the word "hat" for it ... my first point of reference was to the diacritic. Perhaps other people see it as "upsidedown V" for instance.)
 
12:25 PM
@Silverfish: Detexify can be useful. For example, when I draw a hat it suggests \textasciicircum; Googling this command gives "caret/circumflex/hat" as names for the symbol in the first search result.
2
 
 
4 hours later…
4:31 PM
Can anyone point me to a tutorial or something on feature engineering for recurrent neural networks?
 
5:19 PM
@Scortchi Great point! I've never considered using Detexify to find the possible name of a symbol (I've used it extensively for looking up Latex) but it's actually very effective at it.
 
@Glen_b, Phew, that makes me feel a lot better. I had a small panic attack when I got into the article and got completely lost; I definitely need to learn the mathematical side to probability more. Would this kind of level of probability be taught in a graduate stats/biostats degree? I hope it will be, as I would really like to be at this level one day. @MatthewDrury, Glad I could be of help, I'll be scratching my head along with you.
 
5:49 PM
@ChrisC I can't speak for Stat's program, but this kind of probability theory is a common part of post-grad mathematics study. It originated with Kolmogrov, who, I believe, was unsatisfied with the mathematical foundations of conditional probabilities.
Kolmogrov did lots of very important mathematics: en.wikipedia.org/wiki/Andrey_Kolmogorov
 
@MatthewDrury That's really interesting, thanks for the info! If you don't mind me asking, what's your academic background? Did you do a masters or PhD in mathematics?
 
6:19 PM
@ChrisC PhD study in mathematics, Differential Geometry. To my eternal embarrassment, I never actually finished. The economy got bad and I got scared, took a job with the intent of finishing in my free time. Never happened.
 
 
4 hours later…
10:16 PM
@MatthewDrury In my opinion absolutely no embarrassment in making a stable choice :) And also that sounds like essentially the most intense degree a person can pursue, thanks for sharing!
 

« first day (1872 days earlier)      last day (3139 days later) »