pixiedust (version 0.8.3)

glance_foot: Prepare Glance Statistics for pixiedust Table Footer

Description

Retrieves the broom::glance output for a model object and structures it into a table suitable to be placed in the footer. By default, the statistics are displayed in two column-pairings (see Details). This function is not exported but is documented to maintain clarity of its behavior. It is intended for use within dust, but may be useful elsewhere if used with caution.

Usage

glance_foot(fit, col_pairs, total_cols, glance_stats = NULL, byrow = FALSE)

Arguments

fit

A model object with a broom::glance method.

col_pairs

An integer indicating the number of column-pairings for the glance output. This must be less than half the total number of columns, as each column-pairing includes a statistic name and value.

total_cols

The total number of columns in the body of the pixiedust table

glance_stats

A character vector giving the names of the glance statistics to put in the output. When NULL, the default, all of the available statistics are retrieved. In addition to controlling which statistics are printed, this also controls the order in which they are printed.

byrow

A logical, defaulting to FALSE, that indicates if the requested statistics are placed with priority to rows or columns. See Details.

Details

Statistics are placed in column-pairings. Each column pair consists of two columns named stat_name_x and stat_value_x, where x is the integer index of the column pair. The column-pairings are used to allow the user to further customize the output, more-so than pasting the name and value together would allow. With this design, statistics can be rounded differently by applying sprinkles to the resulting table.

The total number of column-pairings must be less than or equal to half the number of total columns. This constraint prevents making glance tables that have more columns than the model table it accompanies.

When the total number of column-parings is strictly less than half the total number of columns, "filler" columns are placed between the column pairings. As much as possible, the filler columns are placed evenly between the column pairings, but when the number of filler columns is unequal between column-pairings, there will be more space placed on the left side. For example, if a table has 7 columns and 3 column-pairings, the order of placement would be column-pair-1, filler, column-pair-2, column-pair-3. Since there was only room for one column of filler, it was placed in the left most fill position.

The byrow arguments acts similarly to the byrow argument in the matrix function, but defaults to FALSE. If four statistics are requested and byrow = FALSE, the left column-pair will have statistics one and two, while the right column-pair will have statistics three and four. If byrow = TRUE, however, the left column-pair will have statistics one and three, while the right column-pair will have statistics two and four.