Skip to content

Organize regression output

4 messages · Francesca PANCOTTO, Michael Dewey, Bert Gunter +1 more

Original
Francesca PANCOTTO
# 
Dear Contributors

I would like to ask some help concerning the automatization process of an analysis, that sounds hard to my knowledge.
I have a list of regression models. 
I call them models=c(ra,rb,rc,rd,re,rf,rg,rh)

I can access the output of each of them using for example, for the first

ra$coefficients

and i obtain

(Intercept)       coeff1          coeff2            age      		gender  
 0.62003033  0.00350807 -0.03817848 -0.01513533 -0.18668972
and I know that ra$coefficients[1] would give me the intercept of this model.

What i need to do is to collect the coefficients of each regression in models, and calculate and place in a table, the following simple summation:


ra						rb	 						rc      ...

intercept					intercept					intercept
intercept+coeff1			intercept+coeff1			intercept+coeff1
intercept+coeff2			intercept+coeff2			intercept+coeff2
intercept+coeff1+coeff2		intercept+coeff1+coeff2		intercept+coeff1+coeff2


The calculations are trivial(I know how to do it in steps) but what is difficult for me is to invent a procedure that organizes the data in an efficient way.

I tried some step , starting with collecting the coefficients but i think I am going the wrong way

calcolati <- list()
for (i in c(ra,rb,rc,rd,re,rf,rg,rh))
{
  calcolati[[i]] <- i$coefficients[1]
}

Thanks for any help you can provide.

f.
----------------------------------
Francesca Pancotto
Web: https://sites.google.com/site/francescapancotto/ <https://sites.google.com/site/francescapancotto/>
Energie: 
http://www.energie.unimore.it/ <http://www.energie.unimore.it/>
----------------------------------
Michael Dewey
# 
Dear Francesca

i usually do this by collecting the models into a list not a vector

model <- list(ra = ra, rb = rb, and so on

and then I use lapply or sapply to process the model

lapply(mode, function(x) coef(x)[1])

or something like that, not tested
On 10/12/2016 07:32, francesca Pancotto wrote:

  
    
  


  


      

    

    

      
      

        


  

        

      Bert Gunter
    

    

      
      #
    

  


  

      
Francesca:

It is hard to know what you might mean by "organize the results in an
efficient way." Some would say that a list of models already does this
for you.

The task you described is fairly simple, so I believe you would
benefit by going through an R tutorial or two to improve your R
programming skills. There are many good ones on the Web. That would
probably be a better long term strategy for you than posting basic
questions here.

Cheers,
Bert




Bert Gunter

"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )


On Fri, Dec 9, 2016 at 11:32 PM, francesca Pancotto

        
<francesca.pancotto at gmail.com> wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Jim Lemon
    

    

      
      #
    

  


  

      
Hi Francesca,
I'm not sure what you are doing here, but try this:

regnames<-paste("r",letters[1:8],sep="")
for(i in 1:8) {
 response<-rnorm(20)
 coef1<-rnorm(20)
 coef2<-rnorm(20)
 age<-sample(20:50,20)
 gender<-sample(c("M","F"),20,TRUE)
 assign(regnames[i],lm(response~coef1+coef2+age+gender))
}
reglist<-list(ra,rb,rc,rd,re,rf,rg,rh)
cumsumcoef<-function(x) return(cumsum(x$coef))
coefsumlist<-as.data.frame(lapply(reglist,cumsumcoef))
names(coefsumlist)<-regnames
coefsumlist

Jim

On Sat, Dec 10, 2016 at 6:32 PM, francesca Pancotto

        
<francesca.pancotto at gmail.com> wrote: