[R] Non linear programming optimization (which package to use?)

[Erwin Kalvelagen]

Jerome Chong <jeromechong <at> gmail.com> writes:

> 
> Greetings,
> 
> What I want to do is to perform regression the data set to get a
> revenue function, and use that revenue function to find the maximum
> point (maybe with some constraints applied).
> The regressed revenue function would most likely be in a quadratic
> form (depending on the price-volume data set), and the constraints
> most likely will be in linear form.
> If I just get the largest product from the data set, it might not give
> me the maximum revenue from the revenue function (assuming not
> violating any constraints).
> To take the constraints into account, won't I be needing non-linear 
programming?
> 
> And if I expect the variable to be an integer value (say price for
> maximized revenue needs to be in integer), I would be requiring mixed
> integer programming.
> 
> Is there a single package can I use to handle all these?
> 
> /Jerome
> 
> 

This optimization problem is called MINLP for Mixed Integer Nonlinear 
Programming. Don't know if there are MINLP solvers for R but they are available 
for other environments (e.g. GAMS).

The first part is regression, that is of course very much in the realm of R.

----------------------------------------------------------------
Erwin Kalvelagen
Amsterdam Optimization Modeling Group
erwin at amsterdamoptimization.com
http://amsterdamoptimization.com