Prev: How Do I Nail Down Chart Width for Printing
Next: Duplicate a chart with relative formulas
From: cloclo on 23 Mar 2010 09:55
Something I have discovered for long but which continues with Excel
2007 (under XP): when you draw a trend line in a scatter or line chart
I generally get either not the best fit (or equation) and/or almost
sytematically the wrong equation associated to this trend line.

Has anybody seen that already and moreover knows why it is like that?
Surprising from a company like Micosoft.
From: Mike Middleton on 23 Mar 2010 15:13
cloclo -

Please provide an example.

In my experience, Excel trend line provides best fit for a linear function
in an XY (Scatter) chart type.

For non-linear functions, the trend line is based on transforms of the
original data, and often you could obtain a better fit using Solver to
search for function parameter values.

For a Line chart type, the trend line uses 1,2,3,... for the X values of the
fitted equation, ignoring any numerical values you may be using as labels
for the horizontal axis.

- Mike
http://www.MikeMiddleton.com


"cloclo" <claude.pluzanski(a)airbus.com> wrote in message
news:24b68b19-1ed6-40b4-938d-341a26337eaa(a)q15g2000yqj.googlegroups.com...
> Something I have discovered for long but which continues with Excel
> 2007 (under XP): when you draw a trend line in a scatter or line chart
> I generally get either not the best fit (or equation) and/or almost
> sytematically the wrong equation associated to this trend line.
>
> Has anybody seen that already and moreover knows why it is like that?
> Surprising from a company like Micosoft.

 | 
Pages: 1
Prev: How Do I Nail Down Chart Width for Printing
Next: Duplicate a chart with relative formulas