[R] GLM FITTED VALUES TABLE

allan clark allan at stats.uct.ac.za
Fri Nov 28 14:01:56 CET 2003


   Hi all

   I have the following generalized linear problem.

   In a study of allergic responses, patients arriving at a clinic in
   Groote Schuur hospital were tested for sensitivity to a number of
   substances. Three of these were moulds: Cladosporium (C), Alternaria
   (F) and Aspergillius (T). Their level of sensitivity was measured on
   the Rast Scale as: 0: not allergic   1: mildly allergic   2 or more:
   allergic

   The data is supplied below. The analysis is fairly straight forward
   and I understand how R solves the problem. I've supplied a copy of the
   code in order to perform the analysis.

   My question is: HOW DOES ONE CONVERT THE OUTPUT (FITTED VALUES)
   SUPPLIED BY R AND DISPLAY THEM IN A CONTINGENCY TABLE?

   allergy<-read.table("c:/a.dat",header=T)
   attach(allergy)
   allergy.fit.main.2int<-glm(y~ .^2,family=poisson, data=allergy)
   fitted(allergy.fit.main.2int)

   The fitted values are:
           1          2          3          4          5
   6          7          8
   666.198041  25.858478  14.943481  60.568127   5.350317   5.081557
   28.233832   3.791205
            9         10         11         12         13
   14         15         16
    10.974962  32.170233   7.068334   1.761433  14.077680   7.039304
   2.883016   7.752087
           17         18         19         20         21
   22         23         24
     5.892362   7.355551  35.631726  14.073188  14.295086   7.354193
   6.610380  11.035427
           25         26         27
     9.014081  12.316433  62.669487

   The data for those interested is as follows:

   > allergy
         t    f    c   y
   1  none none none 671
   2  none none  one  23
   3  none none  two  13
   4  none  one none  60
   5  none  one  one   8
   6  none  one  two   3
   7  none  two none  24
   8  none  two  one   4
   9  none  two  two  15
   10  one none none  31
   11  one none  one   9
   12  one none  two   1
   13  one  one none  14
   14  one  one  one   6
   15  one  one  two   4
   16  one  two none   9
   17  one  two  one   5
   18  one  two  two   7
   19  two none none  32
   20  two none  one  15
   21  two none  two  17
   22  two  one none   8
   23  two  one  one   5
   24  two  one  two  12
   25  two  two none  12
   26  two  two  one  13
   27  two  two  two  59

   Regards
   Allan



More information about the R-help mailing list