[R] summing values by week - based on daily dates - but with somedates missing

[Martyn Byng]

Hi,

Yep, that was what it was doing. For a sum across week, try something like

get.week.flag <- function(dd) {
  ## get weekday from the date in dd and code it as Monday = 1, Tuesday = 2 etc
  idd = factor(weekdays(dd),levels=c("Monday","Tuesday","Wednesday","Thursday","Friday","Saturday","Sunday"))

  ## convert to numeric
  ndd = as.numeric(idd)

  ## flag entries where weekday code gets less (this will flag changes in week)
  wflag = c(FALSE,(ndd[-length(idd)] > ndd[-1]))

  ## cumulative sum to get the week flag
  cumsum(wflag) + 1
}

to get a week flag (this is assuming that your data is sorted by date, if not you'll have to sort it first). If you want the week to start on a different day, just change the ordering of the weekdays in the levels statement.

data.frame(date=myframe$date,day=weekdays(myframe$date),week=get.week.flag(myframe$date))

seems to indicate that the function is doing what it should, so you can then amend the previous code to use get.week.flag instead of weekdays, as in

sum.by.week <- function(ff) {
 by.day <- split(ff$value,get.week.flag(ff$dates))
 lapply(by.day,sum)
}

by.grp <- split(myframe,myframe$group)
lapply(by.grp,sum.by.week)

Martyn

-----Original Message-----
From: Dimitri Liakhovitski [mailto:dimitri.liakhovitski at gmail.com] 
Sent: 30 March 2011 18:03
To: Martyn Byng
Cc: r-help
Subject: Re: [R] summing values by week - based on daily dates - but with somedates missing

Thank you, Martyn.
But it looks like this way we are getting sums by day - i.e., across
all Mondays, all Tuesdays, etc.
Maybe I did not explain well, sorry! The desired output would contain
sums for each WHOLE week - across all days that comprise that week -
Monday through Sunday.
Makes sense?
Dimitri

On Wed, Mar 30, 2011 at 12:53 PM, Martyn Byng <Martyn.Byng at nag.co.uk> wrote:
> Hi,
>
> How about something like:
>
> sum.by.day <- function(ff) {
>  by.day <- split(ff$value,weekdays(ff$dates))
>  lapply(by.day,sum)
> }
>
> by.grp <- split(myframe,myframe$group)
>
> lapply(by.grp,sum.by.day)
>
>
> Martyn
>
> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
> On Behalf Of Dimitri Liakhovitski
> Sent: 30 March 2011 15:23
> To: r-help
> Subject: [R] summing values by week - based on daily dates - but with
> somedates missing
>
> Dear everybody,
>
> I have the following challenge. I have a data set with 2 subgroups,
> dates (days), and corresponding values (see example code below).
> Within each subgroup: I need to aggregate (sum) the values by week -
> for weeks that start on a Monday (for example, 2008-12-29 was a
> Monday).
> I find it difficult because I have missing dates in my data - so that
> sometimes I don't even have the date for some Mondays. So, I can't
> write a proper loop.
> I want my output to look something like this:
> group   dates   value
> group.1 2008-12-29  3.0937
> group.1 2009-01-05  3.8833
> group.1 2009-01-12  1.362
> ...
> group.2 2008-12-29  2.250
> group.2 2009-01-05  1.4057
> group.2 2009-01-12  3.4411
> ...
>
> Thanks a lot for your suggestions! The code is below:
> Dimitri
>
> ### Creating example data set:
> mydates<-rep(seq(as.Date("2008-12-29"), length = 43, by = "day"),2)
> myfactor<-c(rep("group.1",43),rep("group.2",43))
> set.seed(123)
> myvalues<-runif(86,0,1)
> myframe<-data.frame(dates=mydates,group=myfactor,value=myvalues)
> (myframe)
> dim(myframe)
>
> ## Removing same rows (dates) unsystematically:
> set.seed(123)
> removed.group1<-sample(1:43,size=11,replace=F)
> set.seed(456)
> removed.group2<-sample(44:86,size=11,replace=F)
> to.remove<-c(removed.group1,removed.group2);length(to.remove)
> to.remove<-to.remove[order(to.remove)]
> myframe<-myframe[-to.remove,]
> (myframe)
>
>
>
> --
> Dimitri Liakhovitski
> Ninah Consulting
> www.ninah.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.
>
> ________________________________________________________________________
> This e-mail has been scanned for all viruses by Star.
> ________________________________________________________________________
>
> ________________________________________________________________________
> The Numerical Algorithms Group Ltd is a company registered in England
> and Wales with company number 1249803. The registered office is:
> Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.
>
> This e-mail has been scanned for all viruses by Star. The service is
> powered by MessageLabs.
> ________________________________________________________________________
>



-- 
Dimitri Liakhovitski
Ninah Consulting
www.ninah.com

________________________________________________________________________
This e-mail has been scanned for all viruses by Star.\ _...{{dropped:12}}