How to use the "Use weights for each response" option on simbiology model analyzer on a Fit program?

I am trying to figure out how to set up the expression that goes into the ERROR MODEL: "Use weights for each response" box. For example, if I have two responses, how to put more weight on one of them? Is it possible to refer to a response by its name, in case I am using multiple responses?

Any help is much appreciated!

Answer 1

[Torsten]

Did you study the example

"Use Group-Specific Dosing Data for Parameter Estimation with Numeric Weights"

under

https://uk.mathworks.com/help/simbio/ref/fitproblem.html

?

The usual (unweighted) computation tries to fit parameters p1,...,pn to an objective function of the form

sum_i (y_i - y_sim_i(p1,...,pn))^2

where the y_i are your measured data and y_sim_i are the simulated responses while a weighted computation tries to fit parameters p1,...,pn to an objective function of the form

sum_i w_i*(y_i - y_sim_i(p1,...,pn))^2

Thus the number of weights equals the number of responses and a higher value for a weight w_i means that you give measurement y_i more importance during the parameter estimation. Reasons for higher weights can be higher reliability of measurements compared to others or differences in the order of magnitude of units for different responses.

[Rebeca Hannah Oliveira]

Thank you!

[Torsten]

It seems there are more options in SimBiology than just the "usual" weighting (called "Weighted Least Squares for Numerical Weights" on the web page below):

sum_i w_i*(y_i - y_sim_i(p1,...,pn))^2

Here is a list of the different possible expressions that are minimized dependent on the model type chosen:

https://uk.mathworks.com/help/simbio/ref/sbiofit.html#bual8r7