Working with Data.frames in R (Using SAS code to describe what I want)r

I've been mostly working in SAS of late, but not wanting to lose what familiarity with R I have, I'd like to replicate something basic I've done. You'll forgive me if my SAS code isn't perfect, I'm doing this from memory since I don't have SAS at home.

In SAS I have a dataset that roughly is like the following example (. is equivalent of NA in SAS)

A  B
1  1
1  3
0  .
0  1
1  0
0  0

If the dataset above was work.foo then I could do something like the following.

/* create work.bar from dataset work.foo */
data work.bar;
set work.foo;

/* generate a third variable and add it to work.bar */
if a = 0 and b ge 1 then c = 1;
if a = 0 and b = 0  then c = 2;
if a = 1 and b ge 1 then c = 3;
if a = 1 and b = 0  then c = 4;
run;

and I'd get something like

A  B  C
1  1  3
1  3  3
0  .  .
0  1  1
1  0  4
0  0  2

And I could then proc sort by C and then perform various operations using C to create 4 subgroups. For example I could get the means of each group with

proc means noprint data =work.bar; 
by c;
var a b;
output out = work.means mean(a b) = a b;
run;

and I'd get a data of variables by groups called work.means something like:

C  A  B
1  0  1
2  0  0
3  2  2
4  1  0

I think I may also get a . row, but I don't care about that for my purposes.

Now in R. I have the same data set that's been read in properly, but I have no idea how to add a variable to the end (like CC) or how to do an operation on a subgroup (like the b开发者_开发百科y cc command in proc means). Also, I should note that my variables aren't named in any sort of order, but according to what they represent.

I figure if somebody can show me how to do the above, I can generalize it to what I need to do.

Assume your data set is a two-column dataframe called work.foo with variables a and b. Then the following code is one way to do it in R:

work.bar <- work.foo
work.bar$c <- with( (a==0 & b>=1) + 2*(a==0 & b==0) + 3*(a==1 & b>=1) + 
               4*(a==1 & b==0), data=work.foo)
work.mean <- by(work.bar[,1:2], work.bar$c, mean)

An alternative is to use ddply() from the plyr package - you wouldn't even have to create a group variable, necessarily (although that's awfully convenient).

ddply(work.foo, c("a", "b"), function(x) c(mean(x$a, na.rm = TRUE), mean(x$b, na.rm = TRUE))

Of course, if you had the grouping variable, you'd just replace c("a", "b") with "c".

The main advantage in my mind is that plyr functions will return whatever kind of object you like - ddply takes a data frame and gives you one back, dlply would return a list, etc. by() and its *apply brethren usually just give you a list. I think.