Most frequent time given multiple values for time in R

Question

I want to find the average time for each user to play a song given a list of times for song plays for different users.

user  time                 action
A     2013-03-25T14:12:24Z PLAY
B     2013-03-28T14:54:30Z LIKE
C     2013-04-18T18:51:10Z LIKE
D     2013-05-07T18:06:24Z PLAY
B     2013-04-23T12:18:41Z PLAY
D     2013-04-29T12:00:16Z PLAY
A     2013-03-27T12:09:37Z PLAY
A     2013-04-16T18:31:44Z PLAY

I only want to include times where action equals PLAY.

Thanks in advance

Answer 1

The following code returns the average hour (of the day) in which each user plays a song:

DF <- 
read.csv(text=
"user,time,action
A,2013-03-25T14:12:24Z,PLAY
B,2013-03-28T14:54:30Z,LIKE
C,2013-04-18T18:51:10Z,LIKE
D,2013-05-07T18:06:24Z,PLAY
B,2013-04-23T12:18:41Z,PLAY
D,2013-04-29T12:00:16Z,PLAY
A,2013-03-27T12:09:37Z,PLAY
A,2013-04-16T18:31:44Z,PLAY",stringsAsFactors=F)

# filter by PLAY
plays <- DF[DF$action == "PLAY",]

# create means by user
byRes <- 
by(plays, plays$user,
   FUN=function(grp){
        dates <- as.POSIXlt(grp$time,format="%Y-%m-%dT%H:%M:%S", tz = "GMT")
        data.frame(user=grp$user[1],AvgHour=mean(dates$hour))
     })

# put the "by" result into a data.frame
res <- do.call(rbind,byRes)


# result :
> res
  user  AvgHour
A    A 14.66667
B    B 12.00000
D    D 15.00000