[Solved] MCalibration Optimization Objectives  


eugenebraswell
Posts: 12
(@eugenebraswell)
Active Member
Joined: 6 years ago

Hi everyone,

Im back again with a few more questions regarding the MCalibration software. I have received a 1-month trial, and on the first day, I am thoroughly enjoying the software.

My question is pretty simple: What exactly is being optimized when the calibration is run? I see that you can choose various fitness functions, but exactly how are these functions calculated? For example, the coefficient of determination, which I am familiar with, is used in linear regression to measure how well data points fit the predicted regression. Is a similar procedure being used here, with a fit being performed for each of the loading and unloading arms (I am attempting to fit parameters for the Bergstrom-Boyce model)?

Basically, how does each of the fitness functions NMAD, R2, MSD, and max(NMAD) compare the experimental and predicted results? I look forward to responses! Being in academia and hoping to publish results, I just want to be able to explain how I obtained the parameters other than saying I put the data in a magic box and out came some awesome results. :tongue:

Thank you,

Frank

Topic Tags
2 Replies
Jorgen
Posts: 3868
Moderator
(@jorgen)
Member
Joined: 4 months ago

Good question. I just noticed the PolyUMod Users manual does not actually specify those definitions.
The attached document specifies how those values are calculated (see page 11).

-Jorgen

PS. sorry for my slow response...

Topic Tags
2 Replies
eugenebraswell
Posts: 12
(@eugenebraswell)
Active Member
Joined: 6 years ago

Jorgen (If I can call you that =D),

Thanks for the response. This information will be very useful for explanatory and demonstrative purposes!

Thanks,

Frank

Jorgen
Posts: 3868
Moderator
(@jorgen)
Member
Joined: 4 months ago

Good question. I just noticed the PolyUMod Users manual does not actually specify those definitions.
The attached document specifies how those values are calculated (see page 11).

-Jorgen

Share:

Please Login or Register