addMean.R

## set up variables
size <- 36000
numUsers <- 4900
# roughly each user has 7 sessions
numSessions <- (numUsers / 7) - ((numUsers / 7) %% 1)

## create data frame
set.seed(123457)
userIds <- sample.int(numUsers, size=size, replace=TRUE)
ssIds <- sample.int(numSessions, size=size, replace=TRUE)
scores <- sample.int(10, size=size, replace=TRUE)

preDf <- data.frame(User=userIds, Session=ssIds, Score=scores)
preDf$User <- as.factor(preDf$User)

## adding overall mean
# way 1
overallDf1 <- transform(preDf, MeanScore=mean(Score, na.rm=TRUE))

# way 2
require(plyr)
overallDf2 <- mutate(preDf, MeanScore=mean(Score, na.rm=TRUE))

# way 3
require(dplyr)
overallDf3 <- preDf %>%
                  mutate(MeanScore=mean(Score, na.rm=TRUE))

# way 4
require(data.table)
preDt <- data.table(preDf)
setkey(preDt, User)
overallDt <- preDt[,list(User=User
                         ,Session=Session
                         ,Score=Score
                         ,MeanScore=mean(Score, na.rm=T))]

# way 5 - personal preference
overallDf4 <- preDt %>%
      mutate(MeanScore=mean(Score, na.rm=TRUE))

## adding mean by user id
# way 1 - not recommended
require(plyr)
postDf1 <- ddply(preDf
                 ,.(User)
                 ,mutate
                 ,MeanScore=mean(Score, na.rm=TRUE))

# way 2
require(dplyr)
postDf2 <- preDf %>%
            group_by(User) %>%
            mutate(MeanScore=mean(Score, na.rm=TRUE)) %>%
            arrange(User)

# way 3
require(data.table)
preDt <- data.table(preDf)
setkey(preDt, User)
postDt <- preDt[,list(Session=Session
                      ,Score=Score
                      ,MeanScore=mean(Score, na.rm=T))
                ,by=User]

# way 4 - personal preference
postDf3 <- preDt %>%
      group_by(User) %>%
      mutate(MeanScore=mean(Score, na.rm=TRUE)) %>%
      arrange(User)