README.md

In renlund/descripteur: Describe Data Sets

descripteur

The goal of descripteur is to make basic data description easy.

Features

• describe data with arbitrary set of describing functions
• compare arbitrary (possibly overlapping) subsets in different ways with arbitrary set of comparing functions
• keeping 'metadata' that can be utilized for pretty tables (LaTeX)

Data description approch

Describing and comparing

In this package, a description of a data frame (or similar object) consist of a grouping and 2 sets of functions, describers and comparers. The simplest grouping is the trivial grouping where we consider 1 group consisting of all the units, in which case there is nothing to compare and a description is the set of describing functions applied, in turn, to all relevant variables.

If a grouping exists, we can apply comparing functions to the different groups. There are 2 different types of comparisons (comp) available:

• "overall" is when the grouping is fed to some function that makes an overall comparison, like an anova analysis (for arbitrary many groups) or a t-test (in which case there must be exactly 2 groups).
• "across" and "adjacent" are pairwise comparisons of each group with respect to either the first, or the previous, group as given by order of the grouping.

If a grouping exist, this will impact the description (desc) (unless specified elsewise):

• "each" is the describing functions applied to all groups, in turn. This option will be default when considering an "overall" comparison.
• "first" is the describing functions applied only to the first group. This will be default when considerig an "across" or "adjacent" comparison.

Data types and a guide

We do not describe all kinds of data in the same manner. Thus, this package differentiates between different kinds of data types:

• real real, i.e. numeric
• bnry binary, i.e. variables with 2 distinct values
• catg categorical,
• date date (duh!), and
• surv class Surv from the survival package.

This will be determined by a 'guide', which is a data frame that for each variables specifies how it is to be described. Any variable with only 2 distinct values will be regarded as bnry, other text- and factor variables are catg, the classes 'Date' and 'POSIXct' are date and numerical variables with enough distinct values (more than real.tol) are real.

If you do not want to accept the defaults you can create a guide via dtable_guide with non-default arguments, or manipulate the resulting object directly.

More details

More details can be found in the vignette describe-data which functions as the major test for this package.

renlund/descripteur documentation built on Sept. 24, 2023, 7:07 p.m.