Hi,
I downloaded the latest version of asreml which provides residuals and
predicted values in a .yht file. However, how are the residuals
calculated?
For example, I tried a small example of a fixed effects model with a
poisson model with a log link. Attached are the various files. For
example, observation one has parity 1 with nlb=1. The predicted value
is then 0.3185 or 1.375, the latter value is in the .yht file. Asreml
reports the the residual as -0.2378 and it is not clear how this is
computed. Is this the residual directly from Schall's approach?
Thanks in advance,
Bruce
SLOPES FOR LOG(ABS(RES)) on LOG(PV) for Section 1
0.45
*
*
*
*
* *
* *
* * * *
* * * * *
* * * * * **
Min Mean Max-0.45455 0.92519E-16 1.2857 omitting 0 zeros
ASREML [21 Mar 2000] Poisson model
Wed Apr 12 08:45:06 2000 8.00 Mbyte Unix fig
*****************************************************
* ASREML 2000 Residuals now written to .yht file, *
* convergence monitoring of parameters to .res file *
*********************************************** ARG *
QUALIFIERS: !NODISPLAY
Reading fig.dat FREE FORMAT skipping 0 lines
Univariate analysis of nlb
Using 24 records [of 24 read from 24 lines of fig.dat ]
Model term Size Type COL Minimum Mean Maximum #zero #miss
1 animal 1 Covariat 1 1.000 12.50 24.00 0 0
2 parity 3 Factor 2 1 1.8333 3 0 0
3 sire 4 Factor 3 1 2.6667 4 0 0
4 nlb 1 Variate 4 1.000 1.667 4.000 0 0
5 mu 1 Constant Term
Forming 4 equations: 4 dense
Initial updates will be shrunk by factor 0.010
Distribution and link: Poisson; Log Mu=exp(XB) V=Mu
NOTICE: 1 (more) singularities,
LogL= 2.76751 S2= 1.0000 21 df 1.000
LogL= 2.65014 S2= 1.0000 21 df 1.000
LogL= 2.65014 S2= 1.0000 21 df 1.000
LogL= 2.65014 S2= 1.0000 21 df 1.000
Final parameter values 1.0000
Deviance from GLM fit -10.49
Variance heterogenity factor [Deviance/DF] 0.50
Source Model terms Gamma Component Comp/SE % C
Variance 24 21 1.00000 1.00000 0.00 0 F
Analysis of Variance DF F-incr F-adj StndErrDiff
5 mu 1 10.77 1.12
2 parity 2 0.31 0.31 0.4314
Solution Standard Error T-value T-prev
2 parity
2 0.287682 0.369274 0.78
3 0.241162 0.483494 0.50 -0.11
5 mu
4 0.318454 0.301511 1.06
SLOPES FOR LOG(ABS(RES)) on LOG(PV) for Section 1
0.45
Finished: Wed Apr 12 08:45:06 2000 LogL Converged