closest previous year
in [Matlab]
|
Prev: regionprops
Next: regionprops
From: John D'Errico on 1 May 2010 09:55 dpb <none(a)non.net> wrote in message <hrh9j6$v62$1(a)news.eternal-september.org>... > dpb wrote: > ... > > > >> tmp=A(A<B); lowinterpolant = tmp(end) > > lowinterpolant = > > 1992 > > >> > > Better...eliminates temporary > > max(A(A<B)) Also terribly poor for large problems, and still not vectorized. John
From: dpb on 1 May 2010 10:05 John D'Errico wrote: > dpb <none(a)non.net> wrote in message .... >> I didn't see the trick to John's histc() answer--I'm sure there's >> something obvious I'm overlooking. > > Look at the help for histc. > > A = [1990;1992;1996;1997;1998]; > > [junk,index] = histc(1993,A); .... Ah...that's what I wasn't thinking of; for some reason using the array as the edges didn't strike me... I don't disagree on the other comments; as said had a temporary blind spot on histc() --
From: John D'Errico on 1 May 2010 10:30
dpb <none(a)non.net> wrote in message <hrhcih$eh3$1(a)news.eternal-september.org>... > John D'Errico wrote: > > dpb <none(a)non.net> wrote in message > ... > > >> I didn't see the trick to John's histc() answer--I'm sure there's > >> something obvious I'm overlooking. > > > > Look at the help for histc. > > > > A = [1990;1992;1996;1997;1998]; > > > > [junk,index] = histc(1993,A); > ... > > Ah...that's what I wasn't thinking of; for some reason using the array > as the edges didn't strike me... > > I don't disagree on the other comments; as said had a temporary blind > spot on histc() For your reference, histc is the way that one would decide which interval a point falls in when evaluating a spline. There is a second decent solution, that uses a tricky sort. It is also vectorized, and you may occasionally find it in some older spline codes. The only thing that must be watched for is if the point in question falls outside of the bin edges for histc. This is easily trapped though, by one of several schemes. Years ago before histc existed, I put bindex on the FEX, as a solution to this general problem. It is still there, although I imagine it gets little use now except for those people who are still living with version 4 or 5 of matlab. Bindex used a scheme (out of 4 possibilities provided) where it chose a method that was based on the number of bins, the number of points etc. John |