R - get the most common string value (mode) for a given time period

Question

I had hoped to use ddply's mode function to find the most common string for a certain user by time period.

This relates significantly to this question and this question.

Using a data set similar to this:

Data <- data.frame(
    groupname = as.factor(sample(c("red", "green", "blue"), 100, replace = TRUE)),
    timeblock = sample(1:10, 100, replace = TRUE),
    someuser = sample(c("bob", "sally", "sue"), 100, replace = TRUE))

I'd tried:

groupnameagg<- ddply(Data, .(timeblock, groupname, someuser), summarise, groupmode = mode(groupname))

But that's not doing what I had expected. It returns:

> head(groupnameagg$groupname)
[1] "numeric" "numeric" "numeric" "numeric" "numeric" "numeric"

1. How can I find the most commonly occurring groupname by user by timeblock? With a result similar to:

    timeblock   username  mostcommongroupforuser
        1          bob     red
        1          sally   red
        1          sue     green
        2          bob     green
        2          sally   blue
        2          sue     red

2. If groupname is organized by levels, how might I get the highest level present in each timeblock?

Answer 1

Think aggregate should do the trick for both

PART 1

aggregate(Data$groupname,by=list(Data$timeblock,Data$someuser),
     function(x) { 
          ux <- unique(x) 
          ux[which.max(tabulate(match(x, ux)))] })

PART 2

aggregate(Data$groupname,by=list(Data$timeblock,Data$someuser),
     function(x) { 
         levels(Data$groupname)[max(as.numeric(x))] })