[R] How to sum one column in a data frame keyed on other columns

[Marc Schwartz]

On Tue, 2006-12-12 at 15:34 -0800, George Nachman wrote:
> I have a data frame that looks like this:
> 
> url         time somethingirrelevant visits
> www.foo.com 1:00 xxx                 100
> www.foo.com 1:00 yyy                 50
> www.foo.com 2:00 xyz                 25
> www.bar.com 1:00 xxx                 200
> www.bar.com 1:00 zzz                 200
> www.foo.com 2:00 xxx                 500
> 
> I'd like to write some code that takes this as input and outputs
> something like this:
> 
> url         time total_vists
> www.foo.com 1:00 150
> www.foo.com 2:00 525
> www.bar.com 1:00 400
> 
> In other words, I need to calculate the sum of visits for each unique
> tuple of (url,time).
> 
> I can do it with this code, but it's very slow, and doesn't seem like
> the right approach:
> 
> keys = list()
> getkey = function(m,cols,index) { paste(m[index,cols],collapse=",")  }
> for (i in 1:nrow(data)) { keys[[getkey(data,1:2,i)]] = 0 }
> for (i in 1:nrow(data)) { keys[[getkey(data,1:2,i)]] =
> keys[[getkey(data,1:2,i)]] + data[i,4] }
> 
> I'm sure there's a more functional-programming approach to this
> problem! Any ideas?

See ?aggregate

If your dataframe is called 'DF':

> aggregate(DF$visits, list(DF$url, DF$time), sum)
      Group.1 Group.2   x
1 www.bar.com    1:00 400
2 www.foo.com    1:00 150
3 www.foo.com    2:00 525


HTH,

Marc Schwartz