`data.table` in R - creating columns and using functions

Question

I often find myself needing to calculate the lag percentage change of things and was wondering how my current approach could be simplified?

At the moment, i am creating two new columns in my data - in each of the newly created columns i calculated the percentage change from previous year using the function lag . I prefer to keep working in data.table , as my data will be quite large and I for me the the non standard evaluation is much straight forward than in dplyr . Anyway, below is my (example) data.

How can the function be used inside the data.table (do in do not repeat myself that much)?

year <- c(2012, 2013, 2014, 2015)
value <- c (22,33,44,55)
amount <- c(99, 88, 77, 66)

mydata <- cbind(year, value, amount)
mydata <- as.data.table(mydata)

getPctLag(mydata$value)

mydata <- mydata[ , ':=' (value_pct = paste0(round((value/lag(value) - 1) * 100, digits = 3) , " %"),
                          amount_pct = paste0(round((amount/lag(amount) - 1) * 100, digits = 3) , " %"))]

getPctLag <- function(x){lag_pct = paste0(round((x/lag(x) - 1) * 100, digits = 3) , "%")}

Answer 1

You could specify columns to which you want to apply function in .SDcols . Also since you are using data.table it is better to use shift because lag is from dplyr .

library(data.table)

getPctLag <- function(x)  paste(round((x/shift(x) - 1) * 100, digits = 3) , "%")
cols <- c("value", "amount")

mydata[, paste0(cols, "pct") := lapply(.SD, getPctLag), .SDcols = cols]
mydata

#   year value amount value_pct amount_pct
#1: 2012    22     99       NA%        NA%
#2: 2013    33     88       50%   -11.111%
#3: 2014    44     77   33.333%     -12.5%
#4: 2015    55     66       25%   -14.286%