trendline equations are incorrect

I charted a set of coordinates. The polynomial equation
fit extremely well (r square = 0.99) Unless I'm doing
something wrong, (and I don't think I am; I had someone
else look at it as well) the equation is incorrect.

Is this a bug in excel 2002?

On Tue, 16 Sep 2003 16:04:21 -0700, "Fred" wrote:

I charted a set of coordinates. The polynomial equation
fit extremely well (r square = 0.99) Unless I'm doing
something wrong, (and I don't think I am; I had someone
else look at it as well) the equation is incorrect.

Is this a bug in excel 2002?

It depends. Is the formula you are writing about the one displayed on the
chart? If so, it is unlikely to be incorrect. What is more likely is that you
have not formatted the formula to show enough decimal places to be accurate.

Double click on the box that contains the equation. A Format Data Label dialog
box will open and should have a number option. (If it does not, then you
selected the contents of the label, rather than the label). Format the number
to 10 or 15 decimal places and use the resultant formula.


--ron

A safer approach is to use scientific notation with with 14 deciam
places. Without scientific notation, there is no guarantee that the
coefficients are scaled such that any fixed number of decimal places
will be enough.

Jerry

Ron Rosenfeld wrote:

On Tue, 16 Sep 2003 16:04:21 -0700, "Fred" wrote:


I charted a set of coordinates. The polynomial equation
fit extremely well (r square = 0.99) Unless I'm doing
something wrong, (and I don't think I am; I had someone
else look at it as well) the equation is incorrect.

Is this a bug in excel 2002?


It depends. Is the formula you are writing about the one displayed on the
chart? If so, it is unlikely to be incorrect. What is more likely is that you
have not formatted the formula to show enough decimal places to be accurate.

Double click on the box that contains the equation. A Format Data Label dialog
box will open and should have a number option. (If it does not, then you
selected the contents of the label, rather than the label). Format the number
to 10 or 15 decimal places and use the resultant formula.


--ron

Fred -

I charted a set of coordinates. The polynomial equation fit extremely well
(r square = 0.99) Unless I'm doing something wrong, (and I don't think I
am; I had someone else look at it as well) the equation is incorrect. Is
this a bug in excel 2002?

In addition to being sure to display enough significant digits for the
coefficients, be sure to use an XY (Scatter) chart type.

Do not use a Line chart type (in which case the trendline fit uses 1,2,3,...
for the X values).

- Mike Middleton, www.usfca.edu/~middleton