[R] Follow-up Question: data frames; matching/merging

[Gabor Grothendieck]

It works if you use character class rather than factors for your date
strings.  Note last arg to data.frame:

> DF <- data.frame(V1=c(1,1,2,3,1,2),
+ V2=c('2002-03-13','1989-03-10','1988-01-20','1997-05-15','1996-11-18','2000-01-12'),
+ stringsAsFactors = FALSE)

> aggregate(DF[2], DF[1], min)
  V1         V2
1  1 1989-03-10
2  2 1988-01-20
3  3 1997-05-15


On Mon, Feb 8, 2010 at 1:05 PM, Jonathan <jonsleepy at gmail.com> wrote:
> Wow.. thanks for the deluge of responses!
>
> Aggregate seems like the way to go here.
>
> But, suppose that instead of integers in column V2, I actually have
> dates (and instead of keeping the minimum integer, I want to keep the
> earliest date):
>
>> df = data.frame(V1=c(1,1,2,3,1,2),V2=c('2002-03-13','1989-03-10','1988-01-20','1997-05-15','1996-11-18','2000-01-12'));
>> aggregate(df$V2,df['V1'],min) #thanks, Phil
> Error in Summary.factor(c(6L, 2L, 3L), na.rm = FALSE) :
>  min not meaningful for factors
>
> So I checked:
>> min(df$V2)
> Error in Summary.factor(c(6L, 2L, 1L, 4L, 3L, 5L), na.rm = FALSE) :
>  min not meaningful for factors
>
> This can be fixed with:
>> min(as.Date(df$V2))
> [1] "1988-01-20"
>
>
> However, I get a strange response when I try:
>> aggregate(as.Date(df$V2),df['V1'],min)
>  V1    x
> 1  1 7008
> 2  2 6593
> 3  3 9996
>
> What I'm looking for, of course, is:
>> aggregate(as.Date(df$V2),df['V1'],min)
>  V1    V2
> 1  1   1996-11-18
> 2  2   1988-01-20
> 3  3   1997-05-15
>
>
> Any idea how to produce the desired output?
>
> Thanks,
> Jonathan
>
>
>
> On Mon, Feb 8, 2010 at 12:11 PM, Gabor Grothendieck
> <ggrothendieck at gmail.com> wrote:
>> Here are 3 solutions assuming DF contains the data frame:
>>
>>> # 1. aggregate
>>> aggregate(DF[2], DF[1], min)
>>  V1 V2
>> 1  a  2
>> 2  b  9
>> 3  c  4
>>
>>> # 2. aggregate.formula - requires R 2.11.x
>>> aggregate(V2 ~ V1, DF, min)
>>  V1 V2
>> 1  a  2
>> 2  b  9
>> 3  c  4
>>
>>> # 3. SQL using sqldf
>>> library(sqldf)
>>> sqldf("select V1, min(V2) V2 from DF group by V1")
>>  V1 V2
>> 1  a  2
>> 2  b  9
>> 3  c  4
>>
>>> # 4. summaryBy in the doBy package
>>> library(doBy)
>>> summaryBy(V2 ~., DF, FUN = min, keep.names = TRUE)
>>  V1 V2
>> 1  a  2
>> 2  b  9
>> 3  c  4
>>
>> On Mon, Feb 8, 2010 at 11:39 AM, Jonathan <jonsleepy at gmail.com> wrote:
>>> Hi all,
>>>    I'm feeling a little guilty to ask this question, since I've
>>> written a solution using a rather clunky for loop that gets the job
>>> done.  But I'm convinced there must be a faster (and probably more
>>> elegant) way to accomplish what I'm looking to do (perhaps using the
>>> "merge" function?).  I figured somebody out there might've already
>>> figured this out:
>>>
>>> I have a dataframe with two columns (let's call them V1 and V2).  All
>>> rows are unique, although column V1 has several redundant entries.
>>>
>>> Ex:
>>>
>>>     V1     V2
>>> 1    a        3
>>> 2    a        2
>>> 3    b        9
>>> 4    c        4
>>> 5    a        7
>>> 6    b        11
>>>
>>>
>>> What I'd like is to return a dataframe cut down to have only unique
>>> entires in V1.  V2 should contain a vector, for each V1, that is the
>>> minimum of all the possible choices from the set of redundant V1's.
>>>
>>> Example output:
>>>
>>>      V1     V2
>>> 1     a        2
>>> 2     b        9
>>> 3     c        4
>>>
>>>
>>> If somebody could (relatively easily) figure out how to get closer to
>>> a solution, I'd appreciate hearing how.  Also, I'd be interested to
>>> hear how you came upon the answer (so I can get better at searching
>>> the R resources myself).
>>>
>>> Regards,
>>> Jonathan
>>>
>>> ______________________________________________
>>> 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.
>>>
>>
>