No idea, you may want to tell us all your calls and give a reproducible
example as well as properly formatted code that is more readable than
the stuff below.
I guess you omitted some other warnings that tell you about too few
obervations to fit the model (the "fitted probabilities numerically 0 or
1 occurred" is a weak indication that your model may be overspecified).
Best,
Uwe Ligges
On 12.01.2010 08:17, Martin Bulla wrote:> Dear R collegues,
>
> I am new to R and do not understad the warning messages, which appear
> when I am trying to apply type III procedure to my glm:
>
>> inkm$Gamie glm.inkmZ drop1(glm.inkmZ,test="Ch")
> Single term deletions
>
> Model:
> InkMb ~ IncStart + Vol + Gamie + PC1 + PC2 + PC3 + SpotsN + PC1:SpotsN
> Df Deviance AIC LRT Pr(Chi) 9.3879
> 27.388 IncStart 1 14.6274 30.627 5.2395
> 0.0220791 * Vol 1 15.9723 31.972 6.5844 0.0102876 *
> Gamie 1 13.8659 29.866 4.4780 0.0343330 * PC2 1
> 9.6899 25.690 0.3020 0.5826585 PC3 1 10.7326 26.733 1.3447
> 0.2462124 PC1:SpotsN 1 21.6517 37.652 12.2638 0.0004618 ***
> ---
> Signif. codes: 0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1 Warning
> messages:
> 1: In glm.fit(x[, jj, drop = FALSE], y, wt, offset = object$offset, :
> fitted probabilities numerically 0 or 1 occurred
> 2: In glm.fit(x[, jj, drop = FALSE], y, wt, offset = object$offset, :
> fitted probabilities numerically 0 or 1 occurred
>
> Where did PC1 disappiered? What is the meaning of these warning messages?
>
> Similarly, I do not understand where did PCs dissapeared in the
> following calculations:
>> inkm$Gamie glm.inkm0 drop1(glm.inkm0,test="Ch")
> Single term deletions
>
> Model:
> InkMb ~ IncStart + Vol + Gamie + PC1 + PC2 + PC3 + SpotsN + Gamie:PC1 +
> Gamie:PC2 + Gamie:PC3
> Df Deviance AIC LRT Pr(Chi) 18.368
> 40.368 IncStart 1 26.078 46.078 7.7106 0.005490 **
> Vol 1 26.515 46.515 8.1475 0.004312 **
> SpotsN 1 18.663 38.664 0.2958 0.586537 Gamie:PC1 1 20.418
> 40.418 2.0505 0.152155 Gamie:PC2 1 18.460 38.460 0.0919 0.761818
> Gamie:PC3 1 19.832 39.832 1.4647 0.226180 ---
>
> Any suggestions would help me greatly.
> Best regards,
> Martin
>
>
>
> ______________________________________________
> 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.