In krzyslom/lex: Keep track of exponentialized and logarithmized values

Perhaps one is doing dplyr operations on a tibble and does not wish to create additional exp_ or log_ prefixed columns all the time, or maybe has to keep data in not yet decided form in the data base, but wants the columns to be named only once and established in the schema. It is possible to lose track of the current state of the data in the complicated workflow with many different calculations and back and forth log–exp transformations.

lex package was written to help users keep track of data transformation status.

Consider that you have a data series with increasing or decreasing variation e.g. fma::elec.

plot(fma::elec)

One would wish to transform such data so that seasonal variation will be about the same across the whole series.

log_elec <- lex::log(fma::elec)
plot(log_elec)

The information about transformation is held in data's class attribute.

class(log_elec)

One could wish to check for expected data transformation form in crucial moments of the workflow e.g. just before performing addition or multiplication. Common mistake is to mix sum of logharitms with logarithm of sums.

sum_logs <- function(x, y) {
  lex::expect_loged(x) + lex::expect_loged(y)
}

sum_logs(lex::log(1:5), lex::log(21:25))

krzyslom/lex documentation built on May 20, 2019, 10:21 p.m.