Hi,
I would like to use the optimization tool to optimize two parameters in my model(attached), namely beta+AJ.
The .dat format supports only binary data. In case the variable which has to be optimized is a matrix, how can I deal with this data.
I have attached the model below. There are 3 zones. Mij are the migrants coming from zone i, going to zone J, which i want to compare with-optimize to data. The attractivity AJ and beta are the parameters which have to be varied.
optimization with matrix
optimization with matrix
- Attachments
-
- Calibration.mdl
- (2.24 KiB) Downloaded 231 times
-
- Senior Member
- Posts: 1107
- Joined: Wed Mar 12, 2003 2:46 pm
Optimization requires that you specify a numeric payoff - that is a single scalar number. If you have multiple data series you are comparing to this means each needs to be weighted and the payoff definition will sum across the data series and time points. If you are trying to do a policy optimization you will need to weight any entries in a vector or matrix for summation (or geometric averaging or whatever is more appropriate). In the case of policy optimization it is usually easiest to define the payoff inside the model itself.