I would like to log transform several variables in a dataframe and to then add the transformed variables to the dataframe as new variables named using 'logoldname'. What are the best ways of doing these in R efficiently?
data("mtcars")
head(mtcars)
#Log transform - maunally
mtcars$logdisp <- log(mtcars$disp)
mtcars$loghp <- log(mtcars$hp)
mtcars$logwt <- log(mtcars$wt)
mtcars$logqsec <- log(mtcars$qsec)
Here is a tidyverse solution:
# These are the columns with entries you'd like to log-transform
ss <- c("disp", "hp", "wt", "qsec")
mtcars %>%
mutate_at(vars(one_of(ss)), funs(log = log(.))) %>%
rename_at(vars(contains("_log")), funs(paste0("log_", gsub("_log", "", .)))) %>%
select(contains("log_"))
# log_disp log_hp log_wt log_qsec
#1 5.075174 4.700480 0.9631743 2.800933
#2 5.075174 4.700480 1.0560527 2.834389
#3 4.682131 4.532599 0.8415672 2.923699
#4 5.552960 4.700480 1.1678274 2.967333
#5 5.886104 5.164786 1.2354715 2.834389
#6 5.416100 4.653960 1.2412686 3.006672
Explanation: mutate_at selects columns that match ss and applies a log transformation. This generates new columns, named e.g. "disp_log", "hp_log" and so on. We then rename those columns into log_disp, log_hp, etc., and select only the log-transformed columns in the final step.