« first day (2300 days earlier)   

3:32 AM
I'm a little rusty on linear algebra. I'd be grateful if somebody who wasn't could check this answer.
0
A: PCA analysis 100% variance on one principle component

KodiologistThis means all your variables can be written as a linear transformation of a single one of them, which is a pretty extreme case of linear dependence. (Why? Because all the principal components of a dataset together are always a basis for the vector space spanned by the original variables.)

 
 
8 hours later…
11:48 AM
@Kodiologist: Yeap, you are good. MATLAB example: rng(123); svd(repmat(randn(100,1),1,4).*repmat(1:4,100,1)) returns a single non-zero eigenvalue.
 
 
3 hours later…
3:13 PM
@usεr11852 Thanks!
 
 
5 hours later…
7:49 PM
We have some threads that have lot of views but very few upvotes, and often either no, poor, or good but only very brief answers. I guess this means that most of the views are "external", coming from people who cannot vote, i.e. most likely from google searches. Sometimes I am wondering if we should try to curate such threads more attentively, for the benefit of the humanity.
Here is an example that is on the front page now: stats.stackexchange.com/questions/78324 - 9k views, but only 2 upvotes. Gung's answer is good, but short, and only has 1 upvote. Thousands of people are arriving from google to read this thread, but "internally" for us this is clearly an obscure thread. I am sure I've seen similar questions with many thousands of views, 1-2 upvotes and no helpful answers at all.
 
8:19 PM
@Silverfish: Have you considered some professional degrees with significant mathematical content like the CQF? I think I have seen some degrees on Operational Research and Risk Management too. Granted such degrees do entail at least 20-30% regulatory material which maybe is irrelevant for your field. Also given you look towards online education I don't think you need to constraint yourself to UK, for example Penn State
has an Applied Statistics that might a good fit for you. I think I have seen that the University of Illinois had a similar programme.
Both university are rather reputable so maybe that is an option you could also consider.
 
9:14 PM
@amoeba Great point. I think that's worth raising in Meta. I actually noticed this issue before (but can't remember which thread - not that one) and thought it might be worth discussing but I got busy and forgot about it.
@usεr11852 Thanks. That's very helpful. I am sure there is a rather wider range of Masters degrees available if one includes overseas than stick to the UK. Due to somewhat precarious health/pre-existing chronic condition I've generally discounted the US option but that's possibly because I'm coming from a position of ignorance.
(As somebody who is very dependent on the NHS, I have no idea how people get by without it - yet clearly folk do. Much more so than if I'd lived in most other countries, I have almost no practical understanding of the way health insurance schemes and the like work. They are, particularly in their US variant, pretty scary to a typical Brit - they are rather demonised over here!)
Based on what I've heard from people I know in the financial engineering/quant kind of roles, it seems to be very programming-intensive (I view coding as a tool, and sometimes as a chore, but there are certainly people who get rather ... excited ... by it, in a way that I can't, so I'd be happier to leave the field clear for them) and the problems fairly abstract in a way I'd expect to find less interesting.
Of course, it's a wider field than the circle of people I know, so I may be overgeneralising from a small sample. But I did get similar impressions from Quant.SE. I'm more curious about dealing with a range of concrete rather than abstract problems, and a world in which people don't get terribly excited about implementing an algorithm in a way that knocks 5% off the time.
So something like the OR or applied statistics sounds rather more in the correct direction for me.
I don't know much at all about time series (CV has been very educational) but can imagine there being interesting roles in forecasting, for instance.
I haven't noticed this so much in the UK, but abroad in particular there do seem to be a few universities with specialist econometrics or at least time-series-heavy masters that sound like a particularly good platform for forecasting work. But other CVers would know more than me.
 
10:02 PM
@Silverfish: It really depends on your particular health plan, state, etc. During my time in US, I was under of the UC Healthcare System and pretty much everything we automatically sorted even as a foreign national. I fully appreciate your point about healthcare though and that's why I mentioned online programmes. I was under the impression you could attend short periods of times abroad, my bad.
 
10:20 PM
The three (N=3 LOL) bank quants I know came from the same academic research background; SDEs/Stochastics. Again that is my small sample. As mentioned check some of the online programs. Some of them have extremely low-attendance requirements (effectively being like OU). I am just curious though as to what you would like to do with this certificate? Is it just to get a foot through the door in the analytics business?
 
@usεr11852 Not your bad at all - the attendance requirements are actually very interesting/useful.
While the answer to "why bother" is driven largely by intellectual curiosity, it's now so much easier to largely satisfy that without the expenses of university (MOOCs, lecture slides and ebooks online, it's a small point but even getting hold of specialist textbooks was a pain pre-internet-shopping if you didn't have access to an academic library!) that I think a Masters requires a practical justification too!
Probably a view to industry rather than academia.
 
 
1 hour later…
11:37 PM
@Silverfish: "forecasting" is a big field. In the UK, the Lancaster Centre for Forecasting does good work mainly in supply chain forecasting (I'm affiliated), and there are quite a few other supply chain forecasters scattered around - I know some in Cardiff, Bath etc. I think the LSE is a good place to start looking if you are more interested in the econometric side of things.
And yes, I do believe that there are indeed "interesting roles in forecasting" ;-) - one of those roles has been putting food on my family's table for ten years now. Supply chain, though, supermarkets and stuff - not very sexy, but useful.
When people ask me what I do for a living, I tell them I try to ensure that things like this don't happen:
Better Living Through Statistics! Yay!
 
11:55 PM
@StephanKolassa: Nice touch!
 

« first day (2300 days earlier)