Fitting a non-linear model to my data

Question

Jan Hegi 2019-12-1

回答： Jeff Miller 2019-12-1

Hello,

I need to fit a function to my data. I tried it with multiple regression, but it didn't work.

Since the result needs to be 0, if one of the variables is 0, my idea is to do a multiplication of them and find the ideal exponents.

So it would look like this: y = x1^a * x2^b * x3^c * x4^d

Is there any way to find the ideal a, b, c and d if i have the xi and y?

Thanks in advance

Answer 1

Jeff Miller 2019-12-1

Take logs and use multiple regression. That is, fit the model

log(y) = a log(x1) + b log(x2) etc

This will give you the best a, b, c, and d values for a least-squares fit in this log representation