In benjaminrich/lumos: Shed Light on Your Data

knitr::opts_chunk$set(message=FALSE, warning=FALSE, comment='', render=knitr::normal_print)

library(lumos, quietly=TRUE)

The lumos package provides one main function, also called lumos() (but it can be abbreviated simply to l() to avoid typing). The main uses cases of this function are to quickly explore data interactively in the console, or create simple tabular summaries in R markdown documents. Similar to summary(), but aims to be as convenient as possible and produce nicer looking outputs.

If we call lumos() with a data.frame as its sole argument, it outputs a table summarizing the variables, including the columns: variable (its name), label (only present if at least one variable has a label atrribute), class, missing (count) and example (a single value from that variable, typically the first nonmissing value). Let's try it on the Boston housing data from the MASS package:

library(survival)
l(pbc)


library(MASS)
lumos(Boston)

pima <- Pima.tr2
l(pima)


titanic <- as.data.frame(Titanic)
l(titanic)
l(titanic, Survived)
l(titanic, Class, Survived)

When called with data and one other argument, if the argument is categorical outputs a frequency table and if it is continuous outputs a few descriptive statistics (mean, standard deviation, median, min and max). The max option is used to decide if a numeric argument is continuous or categorical.

When called with more than one argument following data, those arguments should all be categorical (.max is ignored in this case). A frequency table is produced for the combinations of the categories, nested from left to right. Percentages are not shown, just counts, and no sorting is done (the categories appear in the order of factor levels).

By default, the function knitr::kable is used to format the output so you get nice looking tables in both the console and in R markdown documents.

If the .gen argument is TRUE, then something different happens. Instead of outputing a table, the function prints code statements into the console: a call to lumos() for each variable in data. The code can be copied from the console back into the script and used to explore the data.frame one variable at a time. This is useful because it saves the need to type the code for each variable.

benjaminrich/lumos documentation built on Oct. 15, 2024, 3:52 a.m.