Programmatically create new columns using column names in a vector with data.table-CodePudding

If we have the following data.table:

x <- data.table(`2021` = rep(10, 3), `2022` = rep(5, 3))

we can create a new column like this:

x[, d := `2022` / `2021` -1]

Now I want to save 2022 and 2021 in a vector like this:

years <- c("2021", "2022")
x[, d := years[2] / years[1] -1]

but of course, this doesn't work. I tried to use eval and as.name but it doesn't work in that specific case. How can I achieve that?

CodePudding user response：

With development version 1.14.3, data.table has gained the new env argument (see the new vignette programming on data.table). So, the recommended way is

library(data.table) # development version 1.14.3 used here
years <- c("2021", "2022")
x[, d := yr2 / yr1 - 1, env = list(yr1 = years[1], yr2 = years[2])][]

x[, d := yrs[[2]] / yrs[[1]] - 1, env = list(yrs = as.list(years))][]

Both return

   2021 2022    d
1:   10    5 -0.5
2:   10    5 -0.5
3:   10    5 -0.5

CodePudding user response：

We could use .SD to subset

x[, d := .SD[[years[2]]]/.SD[[years[1]]] - 1]

CodePudding user response：

Use get when you want to tell the data.table that you are referring to something inside its environment (the dt itself) or higher (see ?get)

x[, d := get(years[2])  / get(years[1]) -1]