Identify columns valid for faceting or splitting

Scans a data frame and returns the names of columns that are appropriate choices for a facet/split selector. A column qualifies when it is categorical (character or factor) and has fewer than max.levels unique values. This keeps facet/split inputs from offering numeric columns or high-cardinality categoricals that would produce an unwieldy number of panels. The modules also use it, through discrete_choices(), to limit their categorical colour, shape and group selectors, where a column of IDs would otherwise ask for one colour or shape per row.

Usage

facet_check(data, max.levels = 50)

Arguments

data: A data frame whose columns are evaluated.
max.levels: A single positive number. Columns with this many or more distinct values are excluded. Defaults to 50.

Value

A character vector of column names suitable for faceting/splitting. Returns character(0) when no column qualifies.

Details

Intended to populate the choices of a facet/split viz_select_input() via update_viz_select() inside a module server, so that only sensible faceting variables are exposed to the user.

A column is considered valid when both of the following hold:

It is categorical: is.character(col) or is.factor(col).
It has fewer than max.levels unique values (NAs excluded).

Numeric columns and categorical columns with max.levels or more distinct values are always excluded.

Author

Jacob Martin

Examples

facet_check(iris)
#> [1] "Species"
facet_check(mtcars)
#> character(0)
# A lower cap drops the 3-level Species column.
facet_check(iris, max.levels = 3)
#> character(0)