[R] glm expand model to more values

[Charles Annis, P.E.]

How many parameters are you trying to estimate?  How many observations do
you have?

What is wrong is that half of your parameter estimates are statistically
meaningless:

dd <- data.frame(a=c(1, 2, 3, 4, 5, 6), b=c(3,  5,  6,  7,  9, 10))

overparameterized.model <- glm(b~poly(a,3),data=dd)

summary(overparameterized.model)


Coefficients:
            Estimate Std. Error t value Pr(>|t|)    

(Intercept)   6.6667     0.1725  38.644 0.000669 ***

poly(a, 3)1   5.7371     0.4226  13.576 0.005382 ** 

poly(a, 3)2  -0.1091     0.4226  -0.258 0.820395    

poly(a, 3)3   0.2236     0.4226   0.529 0.649562  




Charles Annis, P.E.

Charles.Annis at StatisticalEngineering.com
phone: 561-352-9699
eFax:  614-455-3265
http://www.StatisticalEngineering.com
 

-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org] On
Behalf Of Jarek Jasiewicz
Sent: Saturday, January 12, 2008 11:50 AM
To: R-help at r-project.org
Subject: [R] glm expand model to more values

Hi

I have the problem with fitting curve to data with lm and glm. When I 
use polynominal dependiency, fitted values from model are OK, but I 
cannot  recive proper values when I use coefficents to caltulate this.  
Let me present simple example:

I have simple data.frame: (dd)
 a: 1 2 3 4 5 6
 b:  3  5  6  7  9 10

I try to fit it to model:

model=glm(b~poly(a,3),data=dd)
 I have following data fitted to model (as I expected)
 > fitted(model)
        1         2         3         4         5         6
 3.095238  4.738095  6.095238  7.333333  8.619048 10.119048

and coef(model)
(Intercept) poly(a, 3)1 poly(a, 3)2 poly(a, 3)3
  6.6666667   5.7370973  -0.1091089   0.2236068

so when I try to expand the model to other data (simple extrapolation), 
let say: s=seq(1:10,by=1)

I do:
extra=sapply(s,function(x) coef(model) %*% x^(0:3))
and here is result:
[1]  12.51826  19.49328  28.93336  42.18015  60.57528  85.46040 118.17714
 [8] 160.06715 212.47207 276.73354

the data form expanding coefs are completly differnd from fitted

What's going wrong?

Jarek

______________________________________________
R-help at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.