Note: I changed the example from when I first posted. My first example was too simplified to capture the real problem.
I have two data frames
There are several ways to do this (it is R, after all) but I think the most clear is creating an index. We need a function that creates a sequential index (starting at one and ending with the number of observations).
seq_len(3)
> [1] 1 2 3
But we need to calculate this index within each grouping variable (state). For this we can use R's ave
function. It takes a numeric as the first argument, then the grouping factors, and finally the function to be applied in each group.
s1$index <- with(s1,ave(value1,state,FUN=seq_len))
s2$index <- with(s2,ave(value2,state,FUN=seq_len))
(Note the use of with
, which tells R to search for the variables within the environment/dataframe. This is better practice than using s1$value1, s2$value2, etc.)
Now we can simply merge (join) the two data frames (by the variables present in the both data frames: state and index).
merge(s1,s2)
which gives
state index value1 value2
1 IA 1 1 6
2 IA 2 2 7
3 IA 3 3 8
4 IL 1 4 3
5 IL 2 5 4
6 IL 3 6 5
For this to work, there should be the same number of observations by state in each of the data frames.
[Edit: commented the code for clarity.] [Edit: Used seq_len instead of creating a new function as suggested by hadley.]