from Microsoft KB article:
(
http://support.microsoft.com/default...b;en-us;828533)
Microsoft has made extensive changes to the LINEST function to correct
incorrect formulas that are used when the regression line must go through
the origin. The changes also pay more attention to issues that involve
collinear predictor variables. Because of these extensive improvements, this
article focuses more on the improvements and less on instructing users about
how to use LINEST.
Jerry Lewis and David Braden had done a lot of research on the statisical
functions and I believe were major champions/driving forces for changes, som
of which have benn made in xl2003.
--
Regards,
Tom Ogilvy
"Peter T" <peter_t@discussions wrote in message
...
Hi Ron,
Unfortunately, there are certain conditions in which LINEST will not
give
the
correct answer. The formula used in the TRENDLINE box is said to be
more
"robust". I think that means it will give correct answers in situations
where
LINEST gives an incorrect answer.
Tom mentioned similar and I also recall reading about this. However an
observation:
All the macros in this thread (mine with change of number format) return
exactly the same results as LINEST to 14dp, based on Chuck's original data
set. With this set, and all others I've tried that "make sense" with 3rd
order polynomial, suggests that the chart's polynomial trendline uses the
exact same calculation as LINEST.
Chuck's data:
=LINEST({7;10;12;16;26},{5610;11550;16830;22110;26 600}^{1,2,3})
array entered in 4 cells in a row
Would you have an example of data where a polynomial trendline differs
from
LINEST, as in "there are certain conditions ...."
Regards,
Peter T