Test parameters of Wood's model

[Barkamih]

Hi guys 

In this model how to test the params of this model if are they significant or not to fit a lactation curve by using Wood's question? and also what is the best method for this model to use? 

regards 

Ibrahim

proc nlin data = new   plots = fit;
  parms A = 15 B = 0.19 C = -0.0012 ;
  bounds  A B C > 0; 
    by pr  ;
  model  TEST_DAY_MILK_KG = A * Time **b * exp(-C*Time);
  output out = Fit predicted = Pred ;
 run;
QUIT;

[Rick_SAS · Accepted solution]

I don't know what "Wood's method" is, but the ParameterEstimates table includes 95% confidence intervals. If the CIs include 0, then the parameter estimates are not significantly different from 0.

The "best" method is the one that converges fastest for your data. For this model that has one explanatory variable, I see no reason to favor one method over another. I would use the default unless it doesn't converge. 

By the way, your initial guess for C violates the constraint that C > 0.