Dear Allstat,
(Hope this isn't off-topic. If so, I apologise, and please ignore.)
I have been recently thinking about what would be good research areas to
go into. Statistics and related territories like computer science, applied
probability etc. are large. and I'm probably only familiar with a small
portion of the areas.
Anyway, here's a small list of criteria that I would like such an area to
satisfy.
1) Admit of clean, clearly defined problems, preferably in mathematical
terms.
2) Have a clear criteria of success and failure. Ideally I'm looking for
the sort of thing where I can come up with a method or algorithm,
calculate a number, and if it is "better" than the existing number(s) at
least in some situations, then I'm golden. It goes without saying that
such numbers should not be impossibly difficult to come by.
3) Be at least somewhat mathematical. So, for example, not the kind of
thing where you make a model with no good supporting reason or
justification, and hope it has some relationship to "reality", the data
you are modelling, whatever. Also, when I mean mathematical, I mean there
is some interesting, non-trivial math in there, you get to prove things,
and so on.
4) The problem has to be tractable, in the sense that it is not too hard
to make reasonable progress. I don't want to try proving the Riemann
Hypothesis or equivalent.
5) The problem has to be marketable. No point trying to solve problems
nobody cares about.
6) Preferably computationally interesting. I like computing. :-)
Hope all the above doesn't sound too crazy and/or offensive. Not trying to
start a flamewar or step on anyone's toes. I'd be happy to hear if anyone
has thoughts, suggestions, comments about any of this.
Thanks, Faheem.
|