[R] Subseting a data.frame
arun
smartpink111 at yahoo.com
Thu Oct 17 23:33:23 CEST 2013
Hi Bill,
#seq_along() worked in the cases you showed.
ave(seq_along(fac),fac,FUN=length)
#[1] 3 1 3 3
ave(seq_along(num), num, FUN=length)
#[1] 3 1 3 3
ave(seq_along(char), char, FUN=length)
#[1] 3 1 3 3
I thought, there might be some advantages in speed, but they were similar in speed.
set.seed(195)
num1 <- sample(1e3,1e7,replace=TRUE)
system.time(res1 <- ave(integer(length(num1)),num1,FUN=length))
# user system elapsed
#4.148 0.228 4.382
system.time(res2 <- ave(seq_along(num1),num1,FUN=length))
# user system elapsed
# 3.944 0.228 4.181
system.time(res3 <- ave(num1,num1,FUN=length))
# user system elapsed
# 3.740 0.264 4.012
identical(res1,res2)
#[1] TRUE
identical(res2,res3)
#[1] TRUE
A.K.
On Thursday, October 17, 2013 4:34 PM, William Dunlap <wdunlap at tibco.com> wrote:
May I ask why:
count_by_class <- with(dat, ave(numeric(length(basel_asset_class)), basel_asset_class, FUN=length))
should not be more simply done as:
count_by_class <- with(dat, ave(basel_asset_class, basel_asset_class, FUN=length))
The way I did it would work if basel_asset_class were non-numeric.
In ave(x, group, FUN=FUN), FUN's return value should be the same type as x (or
you can get some odd type conversions). E.g.,
> num <- c(2,3,2,2) ; char <- c("Two","Three","Two","Two")
> ave(num, num, FUN=length) # good
[1] 3 1 3 3
> ave(char, char, FUN=length) # bad
[1] "3" "1" "3" "3"
> fac <- factor(char, levels=c("One","Two","Three"))
> ave(fac, fac, FUN=length)
[1] <NA> <NA> <NA> <NA>
Levels: One Two Three
Warning messages:
1: In `[<-.factor`(`*tmp*`, i, value = 0L) :
invalid factor level, NA generated
2: In `[<-.factor`(`*tmp*`, i, value = 3L) :
invalid factor level, NA generated
3: In `[<-.factor`(`*tmp*`, i, value = 1L) :
invalid factor level, NA generated
but x=integer(length(group)) works in all cases:
> ave(integer(length(fac)), fac, FUN=length)
[1] 3 1 3 3
> ave(integer(length(char)), char, FUN=length)
[1] 3 1 3 3
Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com
From: Bert Gunter [mailto:gunter.berton at gene.com]
Sent: Thursday, October 17, 2013 1:06 PM
To: William Dunlap
Cc: Katherine Gobin; r-help at r-project.org
Subject: Re: [R] Subseting a data.frame
May I ask why:
count_by_class <- with(dat, ave(numeric(length(basel_
asset_class)), basel_asset_class, FUN=length))
should not be more simply done as:
count_by_class <- with(dat, ave(basel_asset_class, basel_asset_class, FUN=length))
?
-- Bert
On Thu, Oct 17, 2013 at 12:36 PM, William Dunlap <wdunlap at tibco.com<mailto:wdunlap at tibco.com>> wrote:
> What I need is to select only those records for which there are more than two default
> frequencies (defa_frequency),
Here is one way. There are many others:
> dat <- data.frame( # slightly less trivial example
basel_asset_class=c(4,8,8,8,74,3,74),
defa_frequency=(1:7)/8)
> count_by_class <- with(dat, ave(numeric(length(basel_asset_class)), basel_asset_class, FUN=length))
> cbind(dat, count_by_class) # see what we just computed
basel_asset_class defa_frequency count_by_class
1 4 0.125 1
2 8 0.250 3
3 8 0.375 3
4 8 0.500 3
5 74 0.625 2
6 3 0.750 1
7 74 0.875 2
> mydat[count_by_class>1, ] # I think this is what you are asking for
basel_asset_class defa_frequency
2 8 0.250
3 8 0.375
4 8 0.500
5 74 0.625
7 74 0.875
Bill Dunlap
Spotfire, TIBCO Software
wdunlap tibco.com<http://tibco.com>
> -----Original Message-----
> From: r-help-bounces at r-project.org<mailto:r-help-bounces at r-project.org> [mailto:r-help-bounces at r-project.org<mailto:r-help-bounces at r-project.org>] On Behalf
> Of Katherine Gobin
> Sent: Thursday, October 17, 2013 11:05 AM
> To: Bert Gunter
> Cc: r-help at r-project.org<mailto:r-help at r-project.org>
> Subject: Re: [R] Subseting a data.frame
>
> Correction. (2nd para first three lines)
>
> Pl read following line
>
> What I need is to select only those records for which there are more than two default
> frequencies (defa_frequency), Thus, there is only one default frequency = 0.150 w.r.t
> basel_asset_class = 4 whereas there are default frequencies w.r.t. basel aseet class 4,
>
>
> as
>
> What I need is to select only those records for which there are more than two default
> frequencies (defa_frequency), Thus, there is only one default frequency = 0.150 w.r.t
> basel_asset_class = 4 whereas there are THREE default frequencies w.r.t. basel aseet
> class 8,
>
>
>
> I alpologize for the incovenience.
>
> Regards
>
> KAtherine
>
>
>
>
>
>
>
>
> On , Katherine Gobin <katherine_gobin at yahoo.com<mailto:katherine_gobin at yahoo.com>> wrote:
>
> I am sorry perhaps was not able to put the question properly. I am not looking for the
> subset of the data.frame where the basel_asset_class is > 2. I do agree that would have
> been a basic requirement. Let me try to put the question again.
>
> I have a data frame as
>
> mydat = data.frame(basel_asset_class = c(4, 8, 8 ,8), defa_frequency = c(0.15, 0.07, 0.03,
> 0.001))
>
> # Please note I have changed the basel_asset_class to 4 from 2, to avoid confusion.
>
> > mydat
> basel_asset_class defa_frequency
> 1 4 0.150
> 2 8 0.070
> 3 8 0.030
> 4 8 0.001
>
>
>
> This is just an representative example. In reality, I may have no of basel asset classes. 4, 8
> etc are the IDs can be anything thus I cant hard code it as subset(mydat,
> mydat$basel_asset_class > 2).
>
>
> What I need is to select only those records for which there are more than two default
> frequencies (defa_frequency), Thus, there is only one default frequency = 0.150 w.r.t
> basel_asset_class = 4 whereas there are default frequencies w.r.t. basel aseet class 4,
> similarly there could be another basel asset class having say 5 default frequncies. Thus, I
> need to take subset of the data.frame s.t. the no of corresponding defa_frequencies is
> greater than 2.
>
> The idea is we try to fit exponential curve Y = A exp( BX ) for each of the basel asset
> classes and to estimate values of A and B, mathematically one needs to have at least two
> values of X.
>
> I hope I may be able to express my requirement. Its not that I need the subset of mydat
> s.t. basel asset class is > 2 (now 4 in revised example), but sbuset s.t. no of default
> frequencies is greater than or equal to 2. This 2 is not same as basel asset class 2.
>
> Kindly guide
>
> With warm regards
>
> Katherine Gobin
>
>
>
>
> On Thursday, 17 October 2013 9:33 PM, Bert Gunter <gunter.berton at gene.com<mailto:gunter.berton at gene.com>> wrote:
>
> "Kindly guide" ...
>
> This is a very basic question, so the kindest guide I can give is to read an Introduction to R
> (ships with R) or a R web tutorial of your choice so that you can learn how R works
> instead of posting to this list.
>
> Cheers,
> Bert
>
>
>
>
> On Wed, Oct 16, 2013 at 11:55 PM, Katherine Gobin <katherine_gobin at yahoo.com<mailto:katherine_gobin at yahoo.com>>
> wrote:
>
> Dear Forum,
> >
> >I have a data frame as
> >
> >mydat = data.frame(basel_asset_class = c(2, 8, 8 ,8), defa_frequency = c(0.15, 0.07,
> 0.03, 0.001))
> >
> >> mydat
> > basel_asset_class defa_frequency
> >1 2 0.150
> >2 8 0.070
> >3 8 0.030
> >4 8 0.001
> >
> >
> >I need to get the subset of this data.frame where no of records for the given
> basel_asset_class is > 2, i.e. I need to obtain subset of above data.frame as (since there
> is only 1 record, against basel_asset_class = 2, I want to filter it)
> >
> >> mydat_a
> > basel_asset_class defa_frequency
> >1 8 0.070
> >2 8 0.030
> >3 8 0.001
> >
> >Kindly guide
> >
> >Katherine
> > [[alternative HTML version deleted]]
> >
> >
> >______________________________________________
> >R-help at r-project.org<mailto: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.
> >
> >
>
>
> --
>
> Bert Gunter
> Genentech Nonclinical Biostatistics
>
> (650) 467-7374<tel:%28650%29%20467-7374>
> [[alternative HTML version deleted]]
______________________________________________
R-help at r-project.org<mailto: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.
--
Bert Gunter
Genentech Nonclinical Biostatistics
(650) 467-7374
[[alternative HTML version deleted]]
______________________________________________
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.
More information about the R-help
mailing list