I have a need to create some statistical functions in PostgreSQL that
will take multiple "column" arguments and return multiple values.
A good example would be a least squares fit of a line through a set of
(x,y) points. In this case, I would want the function to accept two
"columns" of data (i.e. x data and y data) as arguments and return two
parameters: the slope and intercept of a line.
The slope is given by
[sum(x^2)*sum(y) - sum(x)*sum*x*y)] / [M*sum(x^2) -(sum(x))^2]
It's not *that* complicated, but it's not that easy to remember either.
And if you're already doing a lot of work to tease out the x and y
points from various tables, it simplifies your query if you can do a
function call.
I can't tell from what have read about user defined functions and user
defined aggregates whether this kind of function is possible.
Bill Eaton