Re: How to create a "summary" data frame

["Hunsicker, Lawrence" <lawrence-hunsicker@uiowa.edu>]

Good morning and Merry Christmas to all:

And thanks to Steve Su, Alan Hochberg, and Chris Barker for help with my question above.

Steve’s solution, using “tapply”, seems the most straight forward, though one can probably get to the same place using “by”.  The problem with “by” is that when its output is combined into a data frame, it is entered as a series of columns, not rows – even when delisted as suggested by Alan.  “tapply” works nicely to give a column vector that works well with “cbind.”  I am pretty sure that one could use “t(by(x))” but I didn’t actually check that out.

One other caution.  One has to be sure that the names column is entered in “cbind” as an integer, not as character.  Else the other columns are coerced to character.  At least one program that works nicely is given below:

test <- tapply(working7$Scr.missing, working7$Center, function(x) sum(x==0)/length(x))

test1 <- tapply(working7$Scr.missing, working7$Center, function(x) length(x))

test2 <- data.frame(cbind(as.integer(names(test)), test, test1))

names(test2) <- c("Center", "PctScrMeas", "NumAtCtr")

Again, thanks to Steve, Alan, and Chris, and Merry Christmas to everyone.

Larry Hunsicker