BookmarkSubscribeRSS Feed
🔒 This topic is solved and locked. Need further help from the community? Please sign in and ask a new question.
marcioadg
Calcite | Level 5

I am estimating nutrient requirement of animals using NLMIXED. When I convert from Imperial to Metric, NLMIXED is estimating different requirement. That does not make sense biologically of course, why I am having this problem? Is there a way to fix it?

1 ACCEPTED SOLUTION

Accepted Solutions
SteveDenham
Jade | Level 19

What model are you using to estimate requirements--a five parameter logistic?  It may be that you are dealing with a scaling/centering problem.  If you express the independent variable in dimensionless units, then this should not occur.

Steve Denham

View solution in original post

6 REPLIES 6
SteveDenham
Jade | Level 19

What model are you using to estimate requirements--a five parameter logistic?  It may be that you are dealing with a scaling/centering problem.  If you express the independent variable in dimensionless units, then this should not occur.

Steve Denham

marcioadg
Calcite | Level 5

I am not sure that is a five parameter logistic. I was using the NLMIXED. And the problem is exactly what you said. Initially I was using 500 grams i.e., now I changed to 0.500 kg and it is exactly the same.

SteveDenham
Jade | Level 19

Oh.  That can happen too, with non-linear models, just a scaling problem.  No, I was thinking of a five parameter logistic dose response curve to various nutrient levels.  What model are you fitting with NLMIXED?  Standard nonlinear growth curve (Gompertz curve)?  (You see, you have hit on my favorite mathematical biology/statistics area, so we may need to go off-line to talk.)

Steve Denham

marcioadg
Calcite | Level 5

Interesting Steve. I am using a 7-point titration, if that is what you mean. Gompertz is an asymptotic model, correct? I am using the broken-line linear and curvilinera (NLMIXED) and a quadratic polynomial model (MIXED). Is my understanding based on the literature that there is not really the 'best' model, there are dozens of types out there. Let me know your thoughts.

SteveDenham
Jade | Level 19

It is my belief (mine, but I have seen others) that the models you choose from should proceed from your knowledge of the physiological processes involved.  There has been enough work done fitting linear regressions or polynomial regressions in this area, and the likelihood of discovering new insights by fitting models of those types, other than as a reference to the literature, is not high.  Broken line/hockey stick etc. models are richer, but physiological processes really don't seem to operate on a basis of stark discontinuities--there is a shift between processes that is smooth and averaged over cells, organs and animals.  A couple of really good reads are J. D. Murray's Mathematical Biology and D. S. Riggs' The Mathematical Aroach to Physiological Problems. Both are older (Murray-1980 (my version is the 1990 corrected second printing), Riggs-1963 (my version is 1967)), but very helpful.

Steve Denham

marcioadg
Calcite | Level 5

Really appreciate your input. I agree. I will check out these references.

sas-innovate-2024.png

Join us for SAS Innovate April 16-19 at the Aria in Las Vegas. Bring the team and save big with our group pricing for a limited time only.

Pre-conference courses and tutorials are filling up fast and are always a sellout. Register today to reserve your seat.

 

Register now!

What is ANOVA?

ANOVA, or Analysis Of Variance, is used to compare the averages or means of two or more populations to better understand how they differ. Watch this tutorial for more.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 6 replies
  • 1529 views
  • 7 likes
  • 2 in conversation