[R] Using lm to estimate a parameter?

[Hans W Borchers]

Uwe Ligges <ligges <at> statistik.tu-dortmund.de> writes:
> 
> On 19.02.2013 11:23, hellen wrote:
> > Hi,
> > I have a data with three variables (X,Y,Z) and I have an equation as
> > Z=X/(1+L*X/Y) where L is a constant which need to be estimated from data.
> > How should I write the formula in lm or is it possible to fit a linear
> > model in this case?
> 
> Neither, it is nonlinear in the parameters. See ?nls or ?optim, for example.

Well, if the Z values are not too small, you can linearize it as

    U = (X Y - Y Z) / Z = L X

and solve it with lm(U ~ X - 1), that is without absolute term.

> Uwe Ligges
> 
> >
> > Thanks!
> > Hallen
> >
>