Constrain of "Linest" function

[Hoi - Excel Dummy]

Anyone can tell me the limitation of Excel function -
"Linest" to do the Linear Regression ?

I got 68 variables and got some 80 equations.
To find out the variables, I use the function Linest but
I got an error "#REF" as my answer

After checking lots of times in the variables and the
functions, I guess it is the limitation of the
Excel that does not allow me to calculate 68 variables.

please to learn the upper constrain of variables allow in
using the Excel function "Linest"

Thanks

 

[Hans D. Jensen]

There is a limit of 50x50 elements in the matrix
inversion routines in Excel, which is used to do the
Linest. There is no problem in having thousands of
equations, but you cannot have more than 50 variables to
solve for.

Hans.

 

[Jerry W. Lewis]

In Excel 2000, XP, and 2003, the limit for MINVERSE is is a 52x52
matrix. LINEST is limited to 17 variables. It is apparently a
hard-coded limit, since MINVERSE can handle larger cases. You are
correct that the number of observations (instead of variables) is
subject only to worksheet limits.

As a practical matter, I would be extremely surprised if the resulting
model from fitting 68 variables was interpretable in a meaningful way,
even if Excel could fit it (and assuming that it is non-singular, which
may be a stretch). The numerical problems alone could be huge. Without
a massive data set (80 observations is hardly massive) much of the fit
would be random noise even with an algebraic solution to the fitting
problem. A more meaningful approach would be to estimate the first few
principal components and try to interpret them.

This could be programmed in Excel, but would be much simpler in a
dedicated statistics package. If you do not have access to one and have
a limited budget, then consider R
www.r-project.org

Jerry