Controlling Matrix Aesthetics

Matrix Groups

• There are 3 types of groups the user can alter within a grob matrix:

  • cells
  • column_names
  • column_headings

• All matrices / data frames inputted will have cells but they don’t have to have column_names, and they won’t have column_headings until the user adds them in after grob_matrix().

grob_matrix()

• To initialize the aesthetic / structure manipulation process, the user must apply grob_matrix() to his/her matrix/data frame.
• Using view_grob() the user can view the current state of the grob matrix before inserting it into a grob-layout and producing the final PDF report.

df = data.frame(x = c(1, 2), y = c(3, 4), z = c(5, 6))

df %>%
  grob_matrix() %>%
  view_grob()

add_aesthetic()

• Aesthetics will be defined as any aspects of the matrix that affect its visual appearance / looks, like colors or font faces.
• add_aesthetic() applies a specific aesthetic to the entire group of the matrix.
• To see a full list of accepted matrix aesthetics, run ?add_aesthetic.

df %>%
  grob_matrix() %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "red",
    group = "cells"
    ) %>%
  view_grob()

df %>%
  grob_matrix() %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "red",
    group = "cells"
    ) %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "blue",
    group = "column_names"
    ) %>%
  view_grob()

add_structure()

• Structures will be defined as any aspects of the matrix that affect its shape, like padding or column widths.
• add_structure() applies a specific structure to the entire matrix, and not just a specific group of the matrix like adding an aesthetic.
• To see a full list of accepted matrix structures, run ?add_structure.
  • The most applicable structure to matrices is column_widths_p, where providing width proportions for each of the columns is the most useful.

df %>%
  grob_matrix() %>%
  add_structure(
    structure = "column_widths_p",
    value = c(3, 1, 1)
    ) %>%
  view_grob()

• Giving a width proportion of 3 to the first column and a width proportion of 1 to the other two means x will be given 3 times the amount of width as y and z.
  • Default width proportions are determined by the width of the elements in each column.

add_column_headings()

• To add column headings to a grob matrix add_column_headings() must be utilized after grob_matrix().

df %>%
  grob_matrix() %>%
  add_column_headings(
    headings = list("C1", "C2"),
    heading_cols = list(c(1, 2), c(3))
    ) %>%
  view_grob()

• Multiple column headings can be added on.

df %>%
  grob_matrix() %>%
  add_column_headings(
    headings = list("C1", "C2"),
    heading_cols = list(c(1, 2), c(3))
    ) %>%
  add_column_headings(
    headings = list("C3", "C4", "C5"),
    heading_cols = list(1, 2, 3)
    ) %>%
  view_grob()

• If a heading_cols is omitted, it will be filled with an empty space.

df %>%
  grob_matrix() %>%
  add_column_headings(
    headings = list("C1"),
    heading_cols = list(c(1, 2))
    ) %>%
  view_grob()

• Once column headings have been initialized, the user can manipulate its aesthetics.

df %>%
  grob_matrix() %>%
  add_column_headings(
    headings = list("C1"),
    heading_cols = list(c(1, 2))
    ) %>%
  add_aesthetic(
    aesthetic = "background_color",
    value = "blue",
    group = "column_headings"
    ) %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "white",
    group = "column_headings"
    ) %>%
  view_grob()

alter_column_names()

• If the user wants to change the displayed column names to something other than the initial column names passed through grob_matrix(), the user should utilize alter_column_names().

df %>%
  grob_matrix() %>%
  alter_column_names(
    column_names = list("C1", "C2", "C3"),
    column_name_cols = list(1, 2, 3)
    ) %>%
  view_grob()

• Very useful for creating column names that span across multiple columns.

df %>%
  grob_matrix() %>%
  alter_column_names(
    column_names = list("GROUP"),
    column_name_cols = list(1:2)
    ) %>%
  view_grob()

alter_at()

• alter_at() is the function that is truly empowers the user to implement cell-by-cell aesthetic / structure customization.
• Inherits the aesthetic / structure and group from the most previous add_aesthetic() / add_structure() / alter_at().
  • Otherwise, the user must specify the specific aesthetic and group he/she wants to alter.
• The user can select specific columns and/or rows to apply a value/function to.
  • Selected rows must be numeric row indices.
  • Selected columns can either be column names or column indices.
• If no columns/rows are specified then all columns/rows will be altered.
• The value, or the .f argument of alter_at() must be a quosure style lambda ~ fun(.).

df %>%
  grob_matrix() %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "blue",
    group = "cells"
    ) %>%
  alter_at(
    ~ "red",
    columns = c("x", "y"),
    rows = 1
    ) %>%
  view_grob()

• Matrix structures can also be altered (like the column width proportions).

df %>%
  grob_matrix() %>%
  add_structure("column_widths_p", 1) %>%
  alter_at(~ 3, columns = 1) %>%
  view_grob()

• The user can also utilize conditional row selection to determine which rows out of the selected rows and columns to alter.

df %>%
  grob_matrix() %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "blue",
    group = "cells"
    ) %>%
  alter_at(
    ~ "red",
    x > 1
    ) %>%
  alter_at(
    ~ "steelblue",
    y < 4,
    aesthetic = "background_color"
    ) %>%
  view_grob()

• The .f argument is meant to be very flexible, as the user can supply a function to apply to the selected cells to output various aesthetics based on their values.

test_function = function(x) ifelse(x > 3, "purple", "blue")

grob_matrix_with_function = df %>%
  grob_matrix() %>%
  add_aesthetic(
    aesthetic = "text_color",
    value = "white",
    group = "cells"
    ) %>%
  alter_at(
    ~ test_function(.),
    aesthetic = "background_color"
    )

grob_matrix_with_function %>% view_grob()

• In instances where the user might already have a nicely formatted matrix (where numeric columns might be converted to character columns), the user can supply a separate, underlying matrix/data frame to apply a function to or apply the conditional row selection to.

formatted_df = apply(df, 2, function(x) paste0("F", x))

grob_matrix_with_new_data = formatted_df %>%
  grob_matrix() %>%
  alter_at(
    ~ test_function(.),
    x > 1,
    data = df,
    aesthetic = "text_color",
    group = "cells"
    )

grob_matrix_with_new_data %>% view_grob()

Inserting into Grob Layout

• Once the user has the grob matrices in the place he/she wants them, he/she can simply insert them into grob_layout() inside a grob_col().

gl = grob_layout(
  grob_row(grob_col(grob_matrix_with_function)),
  grob_row(grob_col(grob_matrix_with_new_data))
  )

gl %>% view_grob(height = 100, width = 100)