[R] Looking for the first observation within the month

[Gabor Grothendieck]

Use the zoo package to represent data like this.

Here time(z) is a vector of the dates and as.yearmon(time(z))
is the year/month of each date.  With FUN=head1, ave picks out the first
date in any month and aggregate then aggregates over all
values in the same year/month choosing the first one.

Lines <- "Date                    Observation

2007-05-23              20
2007-05-22              30
2007-05-21              10

2007-04-10              50
2007-04-09              40
2007-04-07              30

2007-03-05              10
"

library(zoo)

# z <- read.zoo("myfile.dat", header = TRUE)
z <- read.zoo(textConnection(Lines), header = TRUE)

head1 <- function(x, n = 1) head(x, n)
aggregate(z, ave(time(z), as.yearmon(time(z)), FUN = head1), head1)


For more on zoo try:

library(zoo)
vignette("zoo")

and also read the Help Desk article in R News 4/1 about dates.



On 5/27/07, Albert Pang <albert.pang at mac.com> wrote:
> Hi all, I have a simple data frame, first list is a list of dates (in
> "%Y-%m-%d" format) and second list an observation on that particular
> date.  There might not be observations everyday.  Let's just say
> there are no observations on saturdays and sundays.  Now I want to
> select the first observation of every month into a list.  Is there an
> easy way to do that?
>
> Date                    Observation
> ----                    -----------
> 2007-05-23              20
> 2007-05-22              30
> 2007-05-21              10
>
> 2007-04-10              50
> 2007-04-09              40
> 2007-04-07              30
>
> 2007-03-05              10
>
> The result I need is the data frame
>
> 2007-05-21              10
> 2007-04-07              30
> 2007-03-05              10
>
> or I am equally happy with just the vector c(10, 30, 10)
>
> I am new to R and after going through the manuals and the
> documentation I can gather, I have come up with a convoluted way of
> doing it
>
> 1)  I first get the Date into a vector.  (I am articificially
> reproducing this vector below and call it A)
>
>  > A<-c( as.Date("2007-05-23"), as.Date("2007-05-22"), as.Date
> ("2007-05-21"), as.Date("2007-04-10"), as.Date("2007-04-09"), as.Date
> ("2007-04-07"), as.Date("2007-03-05"))
>  > A
> [1] "2007-05-23" "2007-05-22" "2007-05-21" "2007-04-10" "2007-04-09"
> [6] "2007-04-07" "2007-03-05"
>
>
> 2)  use cut with breaks falling on the months
>
>  > B<-cut(A, breaks="month")
>  > B
> [1] 2007-05-01 2007-05-01 2007-05-01 2007-04-01 2007-04-01 2007-04-01
> [7] 2007-03-01
> Levels: 2007-03-01 2007-04-01 2007-05-01
>
>
> 3)  then split to get a list of vectors group by the boundary of the
> date
>
>  > C<-split(A, B)
>  > C
> $`2007-03-01`
> [1] "2007-03-05"
>
> $`2007-04-01`
> [1] "2007-04-10" "2007-04-09" "2007-04-07"
>
> $`2007-05-01`
> [1] "2007-05-23" "2007-05-22" "2007-05-21"
>
>
> 4)  in a for loop I loop through the elements within the list (the
> elements are vectors of dates) with each vector I find the minimum
> and concatentate it to a final vector D
>
>  > D<-numeric()
>  > for ( i in 1:length(C)){ D <- c( D, min(C[[i]]))}
>  > class(D)<-"Date"
>  > D
> [1] "2007-03-05" "2007-04-07" "2007-05-21"
>
> Next with D, I then go back and find out the positions of the
> elements in D within A.  And then use the result as an index vector
> into the vector of observations (which is not shown here)  I feel
> sure I am doing it the stupid way (or the procedural way)
>
> Is there a more declarative way of doing it?  Any pointers will be
> greatly appreciated!
>
> Thanks a lot in advance,
>
> Albert Pang
>
>
>
>
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
>