问题
How would a real R programmer go about writing blocks of code with redundant steps? Here I'm copy/pasting/editing each line, which works fine for this small analysis, but for bigger ones gets unwieldy. In SAS, I'd write macros. Looking for a generating principle in R.
This example has a few typical patterns, like recoding a set of columns that are consecutive and/or have a numbering pattern. Also the recoding logic is just "replace NA with 0 and then add 1" as input to another algorithm that expects positive integers for input variables, here using the car library.
data$rs14_1 <- recode(data$q14_1,"5=6;4=5;3=4;2=3;1=2;0=1;NA=0")
data$rs14_2 <- recode(data$q14_2,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_3 <- recode(data$q14_3,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_4 <- recode(data$q14_4,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_5 <- recode(data$q14_5,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_6 <- recode(data$q14_6,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_7 <- recode(data$q14_7,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_8 <- recode(data$q14_8,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
data$rs14_9 <- recode(data$q14_9,"5=6;4=5;3=4;2=3;1=2;0=1;NA=1")
回答1:
Assuming that the recoding in the first line is supposed to be the same as the rest:
Recode all data columns, create a new data frame as the result:
newdata <- lapply(data,recode,"5=6;4=5;3=4;2=3;1=2;0=1;NA=0")
Set names for the new data frame based on the old one:
names(newdata) <- gsub("^q","rs",names(newdata))
Put them together:
data <- cbind(data,newdata)
But really, instead you should probably be using:
newdata <- data
newdata[is.na(newdata)] <- 0
newdata <- newdata+1
(rather than recode) to do the transformation, followed by the renaming and cbinding steps.
(It would help if you gave a reproducible example.)
来源：https://stackoverflow.com/questions/25650880/tips-for-refactoring-repetitive-code-in-r