-------------------------------------------------------------
Group: DISHES*HEAT 7 0.0310
Class: DISHES*EDLEVEL 23 0.0309
Group: DISHES*EDLEVEL 6 0.0304
Class: GENDER*DISHES 15 0.0300
Group: GENDER*DISHES 6 0.0296
Class: TELIND*DISHES 13 0.0296
Class: MOBILE*DISHES 14 0.0295
Group: MOBILE*DISHES 5 0.0292
Class: DISHES*NUMCARS 21 0.0292
Group: TELIND*DISHES 5 0.0290
Group: DISHES*NUMCARS 7 0.0288
Class: SNGLMOM*DISHES 13 0.0283
Group: SNGLMOM*DISHES 5 0.0279
Class: DISHES 9 0.0271
Group: DISHES 4 0.0268
Var: COATS 1 0.0228
AOV16: FLATWARE 10 0.0221
Var: FLATWARE 1 0.0196 R2 < MINR2
AOV16: RETURN 2 0.0195 R2 < MINR2
Class: NTITLE*RACE 17 0.0148 R2 < MINR2
Group: NTITLE*RACE 6 0.0146 R2 < MINR2
Var: RETURN 1 0.0146 R2 < MINR2
AOV16: WCOAT 10 0.0122 R2 < MINR2
Class: ORIGIN*EDLEVEL 21 0.0118 R2 < MINR2
Additional effects are not listed
SS and R2 Portion for Effects Chosen for the Target
As in the Example 1 analysis, the original input LEISURE appears to be the most important predictor of how much a
customer spends. Notice that the remaining chosen effects are either AOV16 variables or group variables.
SS and R2 portion for Effects chosen for target: AMOUNT
Effect DF R2 SS EMS
-----------------------------------------------------------------------------
Var: LEISURE 1 0.4827 532690709 290726.1
Group: KITCHEN*LUXURY 5 0.1590 175505155 201879.0
AOV16: APPAREL 12 0.1111 122618569 140145.0
AOV16: DOMESTIC 15 0.0415 45758285.2 117548.7
AOV16: WAPPAR 7 0.0122 13463471 110982.1
Group: LUXURY*STATECOD 6 0.0096 10557346.7 105827.7
AOV16: LEISURE 6 0.0074 8168060.98 101889.8
Final ANOVA Table for the Target
The R-square value of 0.8234 is slightly larger than the 0.7810 value obtained from the initial analysis, which did not
include the AOV16 or group variables.
The final ANOVA table for target: AMOUNT
Effect DF R2 SS
------------------------------------------
Model 52 0.8234 908761597
Error 1913 194915205
Total 1965 1103676802
Partial Listing of the SS and R2 portion for Effects Not Chosen for Target
Table
SS and R2 portion for Effects not chosen for target: AMOUNT
Effect DF R2 SS
-------------------------------------------------------------
Group: KITCHEN*STATECOD 7 0.0040 4434939
Var: APPAREL 1 0.0024 2672342
Var: DOMESTIC 1 0.0035 3889555
AOV16: FREQUENT 11 0.0031 3395145
Group: LUXURY*TMKTORD 4 0.0007 811183
Group: KITCHEN*DISHES 4 0.0012 1286988
Var: FREQUENT 1 0.0007 762668
Group: MARITAL*KITCHEN 4 0.0011 1159219
Group: TMKTORD*STATECOD 6 0.0010 1089776
Group: KITCHEN*TMKTORD 5 0.0017 1880850
AOV16: DPM12 11 0.0042 4623270
Group: NTITLE*KITCHEN 6 0.0009 947581
Group: KITCHEN*RACE 5 0.0006 655343
Group: KITCHEN*EDLEVEL 6 0.0010 1084391
Group: LUXURY*DISHES 3 0.0002 258116
Group: LUXURY*RACE 2 0.0004 419452
Group: KITCHEN*ORIGIN 6 0.0010 1117402
Additional effects not listed
Copyright 2000 by SAS Institute Inc., Cary, NC, USA. All rights reserved.
The USEGROUPS option specifies to retain only the reduced group variables
(the original class variables are not included in the model selection process).
Since the NOAOV16 option was not specified, the AOV16 variables that have
an R-square value less than the MINR2 cutoff are used in the final forward
stepwise regression. Because the NOINTER option is omitted, the two-factor
class interactions are also evaluated in the forward stepwise regression.
usegroups