Add+Mult vs Mix ratio residual error model

Phoenix Certara Forums Modelling and Simulation
1

Hello

I’m having trouble figuring out what Add+Mult is with regard to residual error model.

When I look at the model text for Mix Ratio, then that seems to be the additive + multiplicative error model:

observe(CObs = C + CEps * (1 + C * CMixRatio))

Here I clearly have an additive component and a multiplicative component.

What is Add+Mult then?

observe(CObs = C + CEps * sqrt(1 + C^2 * (CMultStdev/sigma())^2))

I was first thinking that sqrt and ^2 was simply to make sure to have a positive number, but it has massive impact on the fit of my model with add+mult being far better. So there must be something in the math I don’t quite understand. That is why I need help and I would like to fully understand the Add+Mult residual error model.

I’m using NLME 1.3, if that makes any difference.

Best regards

Bill

2

please look at my answer here:

http://www.certara.com/forums/topic/741-addmult-error-model-feature/?hl=cmultstdev

and let me know if you still have questions

3

Thanks

After having read you answer over and over again, I think I moved a step closer to grasping the code, but I still don’t fully understand what sigma is in this context.