Weighted least squares linear regression in excel?
 

I want to draw a best fitting straight line through a cloud of data points by least squares linear regression, but I want to weigh the error between the straight line and actual data points for more distant points exponentially less heavily than for the most recent point.

As far as I know, excel doesn't have this function built-in, does anyone have any creative ideas on how to overcome this manco?

Thanks a lot!

Graphing calculator?

If excel doesn't support it, my graphing calc definitely won't. Furthermore, I want to see actual numbers and coefficients so I'm afraid I'm looking for something else. Or isn't that what you mean?

I may have passed all my economics classes and business Calculus II but the terms are all still greek to me.

To make things clearer: TREND and LINEST are functions that can be used to be used for least sq. linear regression, but in the classic way errors at all points in the x axis are given equal weight. (the sum of the squares all these errors is optimised (/minimised) to find the best coefficient)

Reading that brought back some horrid memories of statistical analysis classes back from University.
WG

Any bright ideas?

Try the curve fitting toolbox from MathWorks

matlab rules for technical stuff like that :thumbsup

I have matlab yes, and undoubtedly I can just manually program it, but I was looking to do it in excel.

Bonkerz, I know how to draw a line, but I want to weigh recent points more heavily...