[R] help with merging 2 data frames

Dimitri Liakhovitski dimitri.liakhovitski at gmail.com
Thu Jul 12 01:55:34 CEST 2012


Thanks a lot, everyone for your helpful suggestions!

Eloi, this is very elegant - thank you!
I did not know 2 columns are allowed to have the same names! Always
good to learn something new.
Thanks again!
Dimitri

On Wed, Jul 11, 2012 at 6:58 PM, Mercier Eloi <emercier at chibi.ubc.ca> wrote:
> This should do the trick :
>
> colnames(y)[1:2]=c("a","a")
> y2=rbind(y[,-1], y[,-2]) #duplicating the "y" matrix so the identifiers are
> only in 1 column
> merged = merge(x,y2)
>
> merged
>    a b  d  e1  e2
> 1 aa 1 10 100 101
> 2 aa 2 20 200 201
> 3 ab 1 30 100 101
> 4 ab 2 40 200 201
> 5 ba 1 50 300 301
> 6 ba 2 60 400 401
> 7 bb 1 70 300 301
> 8 bb 2 80 400 401
>
> desired
>    a b  d  e1  e2
> 1 aa 1 10 100 101
> 2 aa 2 20 200 201
> 3 ab 1 30 100 101
> 4 ab 2 40 200 201
> 5 ba 1 50 300 301
> 6 ba 2 60 400 401
> 7 bb 1 70 300 301
> 8 bb 2 80 400 401
>
>  all(merged==desired)
> [1] TRUE
>
> Cheers,
>
> Eloi
>
>
> On 12-07-11 03:50 PM, Dimitri Liakhovitski wrote:
>>
>> Jorge, thank you!
>> that seems to be working, but unfortunately in real life I have
>> thousands of variables (except for a, a2, a3 and b) so that manually
>> selecting columns (as in c(2:4, 8:9)) would be too difficult...
>> Dimitri
>>
>> On Wed, Jul 11, 2012 at 6:36 PM, Jorge I Velez <jorgeivanvelez at gmail.com>
>> wrote:
>>>
>>> Hi Dimitri,
>>>
>>> Try creating a key for "x" and "y" and then merging the result by that
>>> variable:
>>>
>>> x$key <- with(x, paste(a, b, sep = "/"))
>>> y$key <- with(y, paste(a2, b, sep = "/"))
>>> merge(x, y, by = 'key')[, c(2:4, 8:9)]
>>>
>>> HTH,
>>> Jorge.-
>>>
>>>
>>> On Wed, Jul 11, 2012 at 6:28 PM, Dimitri Liakhovitski <> wrote:
>>>>
>>>> Dear R-ers,
>>>>
>>>> I feel I am close, but can't get it quite right.
>>>> Thanks a lot for your help!
>>>>
>>>> Dimitri
>>>>
>>>> # I have 2 data frames:
>>>>
>>>>
>>>>
>>>> x<-data.frame(a=c("aa","aa","ab","ab","ba","ba","bb","bb"),b=c(1:2,1:2,1:2,1:2),d=c(10,20,30,40,50,60,70,80))
>>>>
>>>>
>>>> y<-data.frame(a2=c("aa","aa","ba","ba"),a3=c("ab","ab","bb","bb"),b=c(1:2,1:2),e1=c(100,200,300,400),e2=c(101,201,301,401))
>>>> (x);(y)
>>>>
>>>> # I'd like to merge them so that the result looks like this:
>>>>
>>>>
>>>>
>>>> desired<-data.frame(a=c("aa","aa","ab","ab","ba","ba","bb","bb"),b=c(1:2,1:2,1:2,1:2),d=c(10,20,30,40,50,60,70,80),
>>>>
>>>>
>>>> e1=c(100,200,100,200,300,400,300,400),e2=c(101,201,101,201,301,401,301,401))
>>>> (desired)
>>>>
>>>> # In other words, I want column e1 and e2 entries from data frame y to
>>>> be repeated based on matching of column a from x and columns a2 and
>>>> then a3 from y.
>>>>
>>>> # I am trying step-by-step - first I am using column a2 from data
>>>> frame y for merging:
>>>> out1<-merge(x,y[-2],by.x=c("a","b"),by.y=c("a2","b"),all.x=T,all.y=F)
>>>> (out1)   # looking good - half of the job is done
>>>>
>>>> # Step2 - does not work
>>>>
>>>> # next line produces columns e1 and e2 twice (in real life I have tons
>>>> of columns like e1 and e2):
>>>> merge(out1,y[-1],by.x=c("a","b"),by.y=c("a3","b"),all.x=T,all.y=F)
>>>>
>>>> # next line also doesn't do the job:
>>>>
>>>>
>>>> merge(out1,y[-1],by.x=c("a","b","e1","e2"),by.y=c("a3","b","e1","e2"),all.x=T,all.y=F)
>>>>
>>>> # Finally, I tried this approach:
>>>> out1<-merge(x,y[-2],by.x=c("a","b"),by.y=c("a2","b"),all.x=T,all.y=F)
>>>> out2<-merge(x,y[-1],by.x=c("a","b"),by.y=c("a3","b"),all.x=T,all.y=F)
>>>> (out1); (out2)
>>>>
>>>> # Now I need to merge these 2 - however, the next line doubles the
>>>> number of entries:
>>>> merge(out1,out2,by=names(out1),all.x=T,all.y=T)
>>>>
>>>> --
>>>> Dimitri Liakhovitski
>>>> marketfusionanalytics.com
>>>>
>>>> ______________________________________________
>>>> R-help at r-project.org mailing list
>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>> PLEASE do read the posting guide
>>>> http://www.R-project.org/posting-guide.html
>>>> and provide commented, minimal, self-contained, reproducible code.
>>>
>>>
>>
>>
>
>
> --
> Eloi Mercier
> Bioinformatics PhD Student, UBC
> Paul Pavlidis Lab
> 2185 East Mall
> University of British Columbia
> Vancouver BC V6T1Z4
>



-- 
Dimitri Liakhovitski
marketfusionanalytics.com



More information about the R-help mailing list