Clarification of data transformations
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Clarification of data transformations



Hi,
I would like to recode a variable age at measurement such that the last level 
corresponds to ages 5+ to fit as a factor a model.  Is this possible in ASREML 
without creating a special data set?

I have been trying to use MIN transformations on page 21 of the Manual (dated 
February 8, 2000) says that the !MIN v takes minimum of v and the data value.
However, page 58 Chapter 4 say:
!MIN v is the minimum of the field and the value v (MAX(<field>,v))

My understanding is that ASREML recodes the levels based on the first value 
found and then applied these transformations.  I.E. the first age of dam 
values in order found in the data set are:
4, 3, 5, 6 and 2
So ASREML creates the vector with values    [4 3 5 6 2], 
!MIN  3 then creates the vector with values [4 3 5 5 5].

Thus, v refers to the index of the vector and not the value.  MAX seems to do 
the same but I have yet to understand the SET transformation.

Thanks in advance for any suggestions,
Bruce

The simple data I have been using:
1 3 4 12
1 4 3 12
1 4 5 13
1 4 6 14
2 2 2 13
2 3 3 12
2 4 4 14
2 4 5 12
2 4 6 11
3 2 2 11
3 3 3 12
3 4 4 12
3 4 5 13
3 4 6 15

The .as file:

Test of coding
 animal
 tc    !I
 code   !I !MIN 3
 y
test.dat
y ~ mu code

The .asr file:

 ASREML [ 8 Feb 2000]  Test of coding

 Thu Feb 10 10:16:00 2000   8.00 Mbyte  Unix   test
  Reading test.dat  FREE FORMAT skipping  0  lines
 Univariate analysis of y
 Using       14 records [of      14 read from      14 lines of test.dat
    ]
  Model term      Size Type    COL   Minimum    Mean      Maximum   #zero #miss
   1 animal          1 Covariat  1  1.000      2.071      3.000         0     0
   2 tc              3 Factor    2      1     1.9286          3         0     0
   3 code            3 Factor    3      1     2.3571          3         0     0
   4 y               1 Variate   4  11.00      12.57      15.00         0     0
   5 mu              1 Constant Term
  Forming  4  equations:  4  dense
 Initial updates will be shrunk by factor    0.548
 NOTICE:     1 (more) singularities,
 LogL=-9.75614     S2=  1.4697         11 df    1.000
 Final parameter values                        1.0000

 Source                Model  terms     Gamma     Component    Comp/SE   % C
 Variance                 14     11   1.00000       1.46970       2.35   0 P

 Analysis of Variance       DF     F-incr      F-adj  StndErrDiff
   5 mu                      1    1505.46     327.51
   3 code                    2       0.43       0.43  0.8807

                     Solution       Standard Error    T-value     T-prev
   3 code
                    3  -0.666667       0.989847         -0.67
                    5   0.833333E-01   0.820738          0.10      0.91
   5 mu
                    4    12.6667       0.699928         18.10
 Finished: Thu Feb 10 10:16:02 2000   LogL Converged

The .sln file:

  code                            4           0.000       0.000
  code                            3         -0.6667      0.9898
  code                            5          0.8333E-01  0.8207
  mu                              1           12.67      0.6999
              Residual            1         -0.6667       12.67
              Residual            2          0.1776E-14   12.00
              Residual            3          0.2500       12.75
              Residual            4           1.250       12.75
              Residual            5          0.2500       12.75
              Residual            6          0.1776E-14   12.00
              Residual            7           1.333       12.67
              Residual            8         -0.7500       12.75
              Residual            9          -1.750       12.75
              Residual           10          -1.750       12.75
              Residual           11          0.1776E-14   12.00
              Residual           12         -0.6667       12.67
              Residual           13          0.2500       12.75
              Residual           14           2.250       12.75



--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml