collapse_time_idea.r

# An alternative approach
# First - a quick function for counting unique values that excludes NAs

count.unique <- function(x) { length(unique(x[!is.na(x)])) }

# Compare:
length(unique(NA))
count.unique(NA)


# Then, using plyr to aggregate
collapsed_timelog <- ddply(collapsed_history,
                           .var = c("Account.Name", "Quarter.End", "filing.estimate"),
                           .fun = function(x) {
                           
    # Grab the appropriate subset of timelog
    x_timelog <- subset(timelog, 
                        subset = Account.Name %in% x$Account.Name &
                                 Billable %in% 1 &
                                 Date >= x$Quarter.End &
                                 Date <= x$filing.estimate &
                                 is.na(Date)
    )
    
    # Aggregate using summarise - basically generates a data.frame with jus
    # the variables I name on the 2nd and 3rd lines
    summarise(x_timelog,
              billable_time = sum(Hours),
              concurrent_services = count.unique(x_timelog$Services.ID)
    )

})
              

# Then merging the aggregated results
collapsed_history_time <- merge(x = collapsed_history,
                                y = collapsed_timelog,
                                by = c("Account.Name", "filing.estimate"),
                                all = TRUE
)

Figured out the warning. The subset is conditioning the date with a vector of dates. So, if we say Date >= unique(x$Quarter.End) & Date <= unique(x$filing.estimate), the function has a 1:1 and finishes without any drama. Thanks so much for this.