topic Using PROC NLIN to fit curves for dairy cattle and Proc Gplot to plot predicted values in Statistical Procedures

Using PROC NLIN to fit curves for dairy cattle and Proc Gplot to plot predicted values

houde030 — Fri, 09 Feb 2018 22:17:05 GMT

I am trying to fit lactation curves for different breeds of dairy cattle using Wood's curve with an adjustment for season (s).

The model is:

y(t) = s a t^b e^-c*t

My current code is:

parms a=3 b=0.25 c=-0.0016;
model milk=season*(a*time**(b)*exp(-c*time);
by breed;
output out=pred1 p=predict r=residuals;
run;

Code for plotting predicted values:

proc gplot data=pred1;
plot predict*time=breed;

run; quit;

However when I plot the predicted production values for separate curves for each breed the graph instead plots all 3 breeds by season. How do I get this corrected so I get the graphs for just the breeds and not every season?

Here's the graph I'm getting and an example of the graph I am trying to get:

this is the graph I am getting that is not rightHere's an example of the graph I want

Thank you

Re: Using PROC NLIN to fit curves for dairy cattle and Proc Gplot to plot predicted values

Sascoder — Fri, 09 Feb 2018 22:23:17 GMT

Try using by statement or where clause

proc sort data = pred1

by breed;

proc gplot data=pred1;

by breed;
plot predict*time;

or for more control to plot one breed at a time:

proc gplot data=pred1;

where breed = 'A';
plot predict*time;

Re: Using PROC NLIN to fit curves for dairy cattle and Proc Gplot to plot predicted values

houde030 — Fri, 09 Feb 2018 22:30:39 GMT

The code still graphs different curves for each season instead of one average curve for breed. I'm thinking there is something wrong in my Proc Nlin model.