R help - Nov 2012 - Can you have a by variable in Lag function as in SAS

Hello,

I want to use lag on a time variable but I have to take date into
consideration ie I don't want days to overlap ie: 
I don't want my first time of today to match my last time of yeterday.

In SAS I would use :

data x;
  set y;
  by date tim;
  previous=lag(tim);
  if first.date then
     do;
	    previous=.;
     end;
run;

How can I do something similar in R?  I can't find any examples anywhere.

Thank you all for your help.



--
View this message in context:
http://r.789695.n4.nabble.com/Can-you-have-a-by-variable-in-Lag-function-as-in-SAS-tp4649647.html
Sent from the R help mailing list archive at Nabble.com.

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of ramoss
> Sent: Thursday, November 15, 2012 11:56 AM
> To: r-help at r-project.org
> Subject: [R] Can you have a by variable in Lag function as in SAS
> 
> Hello,
> 
> I want to use lag on a time variable but I have to take date into
> consideration ie I don't want days to overlap ie:
> I don't want my first time of today to match my last time of yeterday.
> 
> In SAS I would use :
> 
> data x;
>   set y;
>   by date tim;
>   previous=lag(tim);
>   if first.date then
>      do;
> 	    previous=.;
>      end;
> run;
> 
> How can I do something similar in R?  I can't find any examples
> anywhere.
> 
> Thank you all for your help.
> 
> 
I  haven't seen a response to this question yet, so here is an approach that
may work for you.  Let's say you have a data frame called dat that contains
your variables date and tim.  Then using the Lag() function from the Hmisc
package you could do something like this

library(Hmisc)
dat$previous <- ave(dat$tim,dat$date,FUN=Lag)


Hope this is helpful,

Dan

Daniel J. Nordlund
Washington State Department of Social and Health Services
Planning, Performance, and Accountability
Research and Data Analysis Division
Olympia, WA 98504-5204

Hi Ramoss,

There are a few solutions to this - probably the best solution involves proper
handling of time series/date objects.
But, for "an" answer try:

# Build a data frame ...
set.seed(123)
someTimes <- sapply(1:5, function(i) sort(round(runif(6, 0, 24))))
myDf <- data.frame(Day = rep(1:5, each = 6), Hour =  as.vector(someTimes))

# Do the lagging ...
do.call("rbind", lapply(split(myDf, myDf$Day), function(df)
transform(df, Lagged = c(NA, df$Hour[-nrow(df)]))))

Or you could use the "Lag" function (not the "lag" function)
from Hmisc:

library(Hmisc)
set.seed(123)
someTimes <- sapply(1:5, function(i) sort(round(runif(6, 0, 24))))
myDf <- data.frame(Day = rep(1:5, each = 6), Hour =  as.vector(someTimes))

# Do the lagging ...
do.call("rbind", lapply(split(myDf, myDf$Day), function(df)
transform(df, Lagged = Lag(df$Hour))))

kind regards,
---------------
Aga Matoga :
amatoga@mango-solutions.com<mailto:amatoga@mango-solutions.com>

http://www.mango-solutions.com<http://www.mango-solutions.com/>, +44
(0)1249 705 450


--

LEGAL NOTICE\ \ This message is intended for the use of ...{{dropped:22}}

Thank you again all responders. Dan your solution was both easy &
miraculous.



--
View this message in context:
http://r.789695.n4.nabble.com/Can-you-have-a-by-variable-in-Lag-function-as-in-SAS-tp4649647p4649773.html
Sent from the R help mailing list archive at Nabble.com.