beta0 value estimation in fitnlm

How to guess the initial value(beta0) for fitnlm(curve fiiting)?

Suppose this is my dataset.

dataset

A B C D

0.25 20 20 5.286

1.00 20 20 5.376

0.25 30 20 5.465

1.00 30 20 5.555

and

D = f(A,B,C)

Answer 1

[Star Strider]

The usual approach is to make a wild guess as to the magnitudes of the parameters, and hope for the best.

A more sophisticated approach is to use the ga (genetic algorithm) function to search the parameter space and come up with a reasonable approximation of the correct parameters. Then use fitnlm to get the most accurate set of parameter estimates, using the ga results as initial parameter estimates.