Greetings:

Is there a PDL module that provides fast, convenient, linear least
squares fitting? I have been poking around the web and email archive,
but can't find any discussion.  I am thinking of a routine that:

* Only fits to y = m * x + b

* Does no matrix inversion, and is correspondingly
  efficient.

* Is called something like
  ($m, $b, $errors_etc) = linleastsq($x,$y)
  or maybe
  ($m, $b, $errors_etc) = linleastsq($x,$y,$weights)

I guess writing it in pure PDL using S_x S_xx, etc. is
very easy.... so it must already exist.

For a workaround, I use fitpoly1d in PDL::Fit::Polynomial.  But this
usually too slow to be practical: In a script that does lots of other
computation and IO, fitpoly1d requires orders of magnitude more time
than the rest of the script.

Thanks,
John


_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to