R help - Jun 2003 - Forward stepwise procedure w/ stepAIC

I'm attempting to select a model using stepAIC.  I want to use a forward
selection procedure.  I have specified a "scope" option, but must not
be
understanding how this works.  My results indicate that the procedure begins
and ends with the "full" model (i.e., all 17 independent
variables)...not
what I expected.  Could someone please point out what I'm not understanding?
My code is below.

 

Thanks.

 

Marc

 
> m1.stepFwd <- stepAIC(m1, direction="forward",
+                       scope=list(upper=~age.refc.fo + dusz.cat.fo +

+                         ed.du.f + emp.ref.f + geodist3.f + has.fone.f +

+                         healthdu.f + help.du.f + inc.ref.f + maj.act.f +

+                         marryref.f + msa.stat.f + regionrf.f + r.e.ref.f +

+                         sex.ref.f + type.psu.f + urb.stat.f, lower=~1))

Start:  AIC= 4752.19 

 response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f +  

    geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f +  

    maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f +  

    sex.ref.f + type.psu.f + urb.stat.f 

 
> 
> m1.stepFwd$anova
Stepwise Model Path 

Analysis of Deviance Table

 

Initial Model:

response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f + 

    geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f + 

    maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f + 

    sex.ref.f + type.psu.f + urb.stat.f

 

Final Model:

response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f + 

    geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f + 

    maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f + 

    sex.ref.f + type.psu.f + urb.stat.f

 

 

  Step Df Deviance Resid. Df Resid. Dev      AIC

1      NA       NA      5311   4660.192 4752.192
>
 

 


	[[alternative HTML version deleted]]

What's "m1"?  Is it a fit to a ~1?  I've used my own
modification of
stepAIC many times preceeded by "m1 <- lm(y~1, data=...)", and it
has
worked fine for me.

hth.  spencer graves

MZodet at ahrq.gov wrote:
> I'm attempting to select a model using stepAIC.  I want to use a
forward
> selection procedure.  I have specified a "scope" option, but must
not be
> understanding how this works.  My results indicate that the procedure
begins
> and ends with the "full" model (i.e., all 17 independent
variables)...not
> what I expected.  Could someone please point out what I'm not
understanding?
> My code is below.
> 
>  
> 
> Thanks.
> 
>  
> 
> Marc
> 
>  
> 
> 
>>m1.stepFwd <- stepAIC(m1, direction="forward",
> 
> 
> +                       scope=list(upper=~age.refc.fo + dusz.cat.fo +
> 
> +                         ed.du.f + emp.ref.f + geodist3.f + has.fone.f +
> 
> +                         healthdu.f + help.du.f + inc.ref.f + maj.act.f +
> 
> +                         marryref.f + msa.stat.f + regionrf.f + r.e.ref.f
+
> 
> +                         sex.ref.f + type.psu.f + urb.stat.f, lower=~1))
> 
> Start:  AIC= 4752.19 
> 
>  response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f +  
> 
>     geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f +  
> 
>     maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f +  
> 
>     sex.ref.f + type.psu.f + urb.stat.f 
> 
>  
> 
> 
> 
>>m1.stepFwd$anova
> 
> 
> Stepwise Model Path 
> 
> Analysis of Deviance Table
> 
>  
> 
> Initial Model:
> 
> response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f + 
> 
>     geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f + 
> 
>     maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f + 
> 
>     sex.ref.f + type.psu.f + urb.stat.f
> 
>  
> 
> Final Model:
> 
> response ~ age.refc.fo + dusz.cat.fo + ed.du.f + emp.ref.f + 
> 
>     geodist3.f + has.fone.f + healthdu.f + help.du.f + inc.ref.f + 
> 
>     maj.act.f + marryref.f + msa.stat.f + regionrf.f + r.e.ref.f + 
> 
>     sex.ref.f + type.psu.f + urb.stat.f
> 
>  
> 
>  
> 
>   Step Df Deviance Resid. Df Resid. Dev      AIC
> 
> 1      NA       NA      5311   4660.192 4752.192
> 
> 
> 
>  
> 
>  
> 
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help

It starts with the initial model you gave it, which contains all your 
explanatory variables already.  You presumably wanted to use a simpler 
model to start with.  As in:

Arguments:

  object: an object representing a model of an appropriate class. This
          is used as the initial model in the stepwise search.


On Wed, 18 Jun 2003 MZodet at ahrq.gov wrote:
> I'm attempting to select a model using stepAIC.  I want to use a
forward
> selection procedure.  I have specified a "scope" option, but must
not be
> understanding how this works.  My results indicate that the procedure
begins
> and ends with the "full" model (i.e., all 17 independent
variables)...not
> what I expected.  Could someone please point out what I'm not
understanding?
[output snipped]

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595