Curve fitting with Excel

[ajo...@freeuk.com]

Hi

Just a question:

Is it possible to use Excel for non-linear regressions to fit dose
response curves to get IC50 values.... and then to constrain "Tops"
and "Bottoms" with in certain limits?

If so, could you tell me how?

Many thanks

Andrew Johnston
Newport
Fife
Scotland.

[Jerry W. Lewis]

You can use Solver to approximately minimize deviations from the model.
Solver also accepts constraints.

Minimizing the sum of squared deviations would give you the Least
Squares fit. Minimizing the sum of absolute deviations would give you
the L1 estimate. Minimizing the maximum absolute deviation would give
you the Minimax estimate, etc.

Since you are interested in dose response curves, I presume that you are
doing somthing like a 4-parameter logistic curve which truly is
non-linear (but can be linearized if you know the asymptotic values).

In general don't assume that a model is non-linear just because the
graph of the model is curved. A "linear model" is a model that is
linear in the unknowns (such as a polynomial), regardless of what the
known values are. Linear models (including polynomials) can be fit with
LINEST().

Jerry