Merging data frames without duplicating rows

Question

I would like to merge two data frames, but do not want to duplicate rows if there is more than one match. Instead I would like to sum the observations on that day.

Answer 1

I'd suggest you merge them and then aggregate them (essentially perform a SUM for each unique Date).

df <- merge(z.days,obs.days, by.x="Date", by.y="Date", all.x=TRUE)
        Date Count
1 2012-01-01    NA
2 2012-01-02     1
3 2012-01-03     1
4 2012-01-03     1
5 2012-01-04    NA

Now to do the merge you could use aggregate:

df2 <- aggregate(df$Count,list(df$Date),sum)
     Group.1  x
1 2012-01-01 NA
2 2012-01-02  1
3 2012-01-03  2
4 2012-01-04 NA
names(df2)<-names(df)

BUT I'd recommend package plyr, which is awesome! In particular, function ddply.

library(plyr)
ddply(df,.(Date),function(x) data.frame(Date=x$Date[1],Count=sum(x$Count)))
        Date Count
1 2012-01-01    NA
2 2012-01-02     1
3 2012-01-03     2
4 2012-01-04    NA

The command ddply(df,.(Date),FUN) essentially does:

for each date in unique(df$Date):
    add to output dataframe FUN( df[df$Date==date,] )

So the function I've provided creates a data frame of one row with columns Date and Count, being the sum of all counts for that date.