rtables (version 0.6.17)

split_rows_by: Add rows according to levels of a variable

Description

Add rows according to levels of a variable

Usage

split_rows_by(
  lyt,
  var,
  labels_var = var,
  split_label = var,
  split_fun = NULL,
  parent_name = var,
  format = NULL,
  na_str = NA_character_,
  nested = TRUE,
  at_sibling = NULL,
  child_labels = c("default", "visible", "hidden"),
  label_pos = if (!is.null(at_sibling)) "visible" else "default",
  indent_mod = 0L,
  page_by = FALSE,
  page_prefix = split_label,
  section_div = NA_character_
)

Value

A PreDataTableLayouts object suitable for passing to further layouting functions, and to build_table().

Arguments

lyt

(PreDataTableLayouts)
layout object pre-data used for tabulation.

var

(string)
variable name.

labels_var

(string)
name of variable containing labels to be displayed for the values of var.

split_label

(string)
label to be associated with the table generated by the split. Not to be confused with labels assigned to each child (which are based on the data and type of split during tabulation).

split_fun

(function or NULL)
custom splitting function. See custom_split_funs.

parent_name

(character(1))
Name to assign to the table corresponding to the split or group of sibling analyses, for split_rows_by* and analyze* when analyzing more than one variable, respectively. Ignored when analyzing a single variable.

format

(string, function, or list)
format associated with this split. Formats can be declared via strings ("xx.x") or function. In cases such as analyze calls, they can be character vectors or lists of functions. See formatters::list_valid_format_labels() for a list of all available format strings.

na_str

(string)
string that should be displayed when the value of x is missing. Defaults to "NA".

nested

(logical)
whether this layout instruction should be applied within the existing layout structure if possible (TRUE, the default) or as a new top-level element (FALSE). Ignored if it would nest a split underneath analyses, which is not allowed.

at_sibling

(character(1) or NULL)
If non-null, a preceding split or analyze to anchor this instruction to as a direct sibling. Cannot select an instruction that is downstream of a point where a previously used anchor (See Nesting Anchor Resolution for details).

child_labels

(string)
the display behavior for the labels (i.e. label rows) of the children of this split. Accepts "default", "visible", and "hidden". Defaults to "default" which flags the label row as visible only if the child has 0 content rows.

label_pos

(string)
location where the variable label should be displayed. Accepts "hidden" (default for non-analyze row splits), "visible", "topleft", and "default" (for analyze splits only). For analyze calls, "default" indicates that the variable should be visible if and only if multiple variables are analyzed at the same level of nesting.

indent_mod

(numeric)
modifier for the default indent position for the structure created by this function (subtable, content table, or row) and all of that structure's children. Defaults to 0, which corresponds to the unmodified default behavior.

page_by

(flag)
whether pagination should be forced between different children resulting from this split. An error will occur if the selected split does not contain at least one value that is not NA.

page_prefix

(string)
prefix to be appended with the split value when forcing pagination between the children of a split/table.

section_div

(string)
string which should be repeated as a section divider after each group defined by this split instruction, or NA_character_ (the default) for no section divider.

Nesting Anchor Resolution

When nested is TRUE, at_sibling allows you to set a nesting anchor that your new split_rows_by* or analyze* directive should be placed as a sibling to. The lookup for this anchor occurs only in the currently active top-level nesting stack, meaning the directives that have occurred since the last split or analysis with nested == FALSE.

Furthermore, resolution occurs against the first element of each arm of a branching point caused by any previous uses of at_sibling but only descends into the last arm.

So for example if our previous layout was generated via:

lyt <- basic_table() |>
  split_rows_by("SEX") |>
  analyze("AGE") |>
  split_rows_by("BMRKR2", nested = FALSE) |>
  split_rows_by("RACE") |>
  analyze("AGE") |>
  split_rows_by("SEX", at_sibling = "RACE") |>
  analyze("BMRKR1")

The eligible anchor points would be "BMRKR2", "RACE", "SEX" and "BMRKR1". "AGE" is masked by the branching caused by anchoring our SEX split on RACE.

Finally, while at_sibling does support de-duplication of "<name>[i]" anchors, it does so within the set of available anchors, which can be counter-intuitive. It is strongly suggested that the parent_name and table_names argument(s) of split_rows_by* and analyze be used to prevent the need for this. at_sibling will resolve to table names overridden in this manner.

Custom Splitting Function Details

User-defined custom split functions can perform any type of computation on the incoming data provided that they meet the requirements for generating "splits" of the incoming data based on the split object.

Split functions are functions that accept:

df

a data.frame of incoming data to be split.

spl

a Split object. This is largely an internal detail custom functions will not need to worry about, but obj_name(spl), for example, will give the name of the split as it will appear in paths in the resulting table.

vals

any pre-calculated values. If given non-NULL values, the values returned should match these. Should be NULL in most cases and can usually be ignored.

labels

any pre-calculated value labels. Same as above for values.

trim

if TRUE, resulting splits that are empty are removed.

(optional) .spl_context

a data.frame describing previously performed splits which collectively arrived at df.

The function must then output a named list with the following elements:

values

the vector of all values corresponding to the splits of df.

datasplit

a list of data.frames representing the groupings of the actual observations from df.

labels

a character vector giving a string label for each value listed in the values element above.

(optional) extras

if present, extra arguments are to be passed to summary and analysis functions whenever they are executed on the corresponding element of datasplit or a subset thereof.

One way to generate custom splitting functions is to wrap existing split functions and modify either the incoming data before they are called or their outputs.

Author

Gabriel Becker

Examples

lyt <- basic_table() |>
  split_cols_by("ARM") |>
  split_rows_by("RACE", split_fun = drop_split_levels) |>
  analyze("AGE", mean, var_labels = "Age", format = "xx.xx")

tbl <- build_table(lyt, DM)
tbl

lyt2 <- basic_table() |>
  split_cols_by("ARM") |>
  split_rows_by("RACE") |>
  analyze("AGE", mean, var_labels = "Age", format = "xx.xx")

tbl2 <- build_table(lyt2, DM)
tbl2

lyt3 <- basic_table() |>
  split_cols_by("ARM") |>
  split_cols_by("SEX") |>
  summarize_row_groups(label_fstr = "Overall (N)") |>
  split_rows_by("RACE",
    split_label = "Ethnicity", labels_var = "ethn_lab",
    split_fun = drop_split_levels
  ) |>
  summarize_row_groups("RACE", label_fstr = "%s (n)") |>
  analyze("AGE", var_labels = "Age", afun = mean, format = "xx.xx")

lyt3

library(dplyr)

DM2 <- DM |>
  filter(SEX %in% c("M", "F")) |>
  mutate(
    SEX = droplevels(SEX),
    gender_lab = c(
      "F" = "Female", "M" = "Male",
      "U" = "Unknown",
      "UNDIFFERENTIATED" = "Undifferentiated"
    )[SEX],
    ethn_lab = c(
      "ASIAN" = "Asian",
      "BLACK OR AFRICAN AMERICAN" = "Black or African American",
      "WHITE" = "White",
      "AMERICAN INDIAN OR ALASKA NATIVE" = "American Indian or Alaska Native",
      "MULTIPLE" = "Multiple",
      "NATIVE HAWAIIAN OR OTHER PACIFIC ISLANDER" =
        "Native Hawaiian or Other Pacific Islander",
      "OTHER" = "Other", "UNKNOWN" = "Unknown"
    )[RACE]
  )

tbl3 <- build_table(lyt3, DM2)
tbl3