[R] trouble with extraction/interpretation of variance structure para meters from a model built using gnls and varConstPower

Rand, Hugh rand at amgen.com
Mon Jan 9 16:37:51 CET 2006

I have been using gnls with the weights argument (and varConstPower) to
specify a variance structure for curve fits. In attempting to extract the
parameters for the variance model I am seeing results I don't understand.
When I simply display the model (or use "summary" on the model), I get what
seem like reasonable values for both "power" and "const". When I actually
try to extract the values, I get the same number for the "power", but a
different (and less sensible) value for "const".

The simplest example I can come up with that shows the problem is as

#Set up data

x    = c(0,0,0,1,1,1,2,2,2,3,3,3,4,4,4,5,5,5,6,6,6,7,7,7);
y0   = c(.1,.1,.1, .5,.5,.5, 1,1,1, 2,2,2, 4,4,4, 7,7,7, 9,9,9, 10,10,10);
yp   = c(0,.03,.05, 0,.05,.01, 0,.07,.03, .1,0,.2, .2,.1,0, .3,0,.1,
0,.3,.4, .3,.5,0);
y    = y0 + 4*yp;
data = data.frame(x=x,y=y);

#Run model

model3 = try(gnls(y ~

#Examine results

model3;                                         #const = .6551298,   power =
summary(model3);                                #const = .6551298,   power =

coef(model3$modelStruct$varStruct)["power"];    #                    power =
coef(model3$modelStruct$varStruct)["const"];    #const = -0.4229219 
coef.varFunc(model3$modelStruct$varStruct);     #R can't seem to find this
function, Splus can

Any advice on what I am doing wrong would be appreciated.

Hugh Rand
Senior Scientist
rand at amgen.com

More information about the R-help mailing list