How to efficiently divide successor by predecessor in each column of a dataframe

Question

I have a dataframe myDF created like this:

a <- 1:4
b <- seq(3, 16, length=4)
myDF <- data.frame(a=a, b=b)

which looks like this:

  a         b
1 1  3.000000
2 2  7.333333
3 3 11.666667
4 4 16.000000

Now I want to divide subsequently predecessor and successor in each column, add the results to the existing dataframe, replace the one missing value in each column by NA and add new column names. For the example above, my desired outcome looks like this:

  a         b     amod     bmod
1 1  3.000000       NA       NA
2 2  7.333333 2.000000 2.444444
3 3 11.666667 1.500000 1.590909
4 4 16.000000 1.333333 1.371429

So, in column a 2 is divided by 1, 3 is divided by 2, and 4 is divided by 3 and the results are stored in amod.

The way I do it now is like this:

divStuff <-function(aCol){

  newCol <- aCol[2:length(aCol)]/aCol[1:length(aCol) - 1]
  newCol <- c(NA, newCol)

  return(newCol)
}
newDF <- data.frame(lapply(myDF, divStuff))
names(newDF) <- paste(names(myDF), "mod", sep="")
endDF <- cbind(myDF, newDF)

I wrote a function divStuff which does the division and then call lapply which applies this function to each column of the data frame.

Now I am wondering whether that is the way to do it or whether there is a smarter way on doing such kind of operations which would e.g. avoid the cbind call or does the cbind in a way which avoids the line newCol <- c(NA, newCol) by adding a NA automatically. I did not find a nice way, all solutions for that looks similar to this one.

Answer 1

Here's a quick data.table version (using the devel version on GH)

library(data.table) ## V 1.9.5
setDT(myDF)[, paste0(names(myDF), "mod") := lapply(.SD, function(x) x/shift(x))]
#    a         b     amod     bmod
# 1: 1  3.000000       NA       NA
# 2: 2  7.333333 2.000000 2.444444
# 3: 3 11.666667 1.500000 1.590909
# 4: 4 16.000000 1.333333 1.371429

Or similarly with dplyr though you may want to play around with the column names (this is due a bug(?) in mutate_each when it drops the original columns and doesn't rename the resulting ones when given a single function)

library(dplyr)
myDF %>% 
  mutate_each(funs(./lag(.))) %>%
  cbind(myDF, .)
#   a         b        a        b
# 1 1  3.000000       NA       NA
# 2 2  7.333333 2.000000 2.444444
# 3 3 11.666667 1.500000 1.590909
# 4 4 16.000000 1.333333 1.371429

Answer 2

With base R:

myDF[,paste0(names(myDF), "mod")] <- sapply(myDF, function(x) c(NA, x[-1]/head(x,-1)))
#  a         b     amod     bmod
#1 1  3.000000       NA       NA
#2 2  7.333333 2.000000 2.444444
#3 3 11.666667 1.500000 1.590909
#4 4 16.000000 1.333333 1.371429