build_table {rtables}

Create a table from a layout and data

Description

Layouts are used to describe a table pre-data. build_table is used to create a table using a layout and a dataset.

Usage

build_table(
  lyt,
  df,
  alt_counts_df = NULL,
  col_counts = NULL,
  col_total = if (is.null(alt_counts_df)) nrow(df) else nrow(alt_counts_df),
  topleft = NULL,
  hsep = default_hsep(),
  ...
)

Arguments

lyt

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

df

(data.frame or tibble)
dataset.

alt_counts_df

(data.frame or tibble)
alternative full dataset the rtables framework will use only when calculating column counts.

col_counts

(numeric or NULL)
if non-NULL, column counts for leaf-columns only which override those calculated automatically during tabulation. Must specify "counts" for all leaf-columns if non-NULL. NA elements will be replaced with the automatically calculated counts. Turns on display of leaf-column counts when non-NULL.

col_total

(integer(1))
the total observations across all columns. Defaults to nrow(df).

topleft

(character)
override values for the "top left" material to be displayed during printing.

hsep

(string)
set of characters to be repeated as the separator between the header and body of the table when rendered as text. Defaults to a connected horizontal line (unicode 2014) in locals that use a UTF charset, and to - elsewhere (with a once per session warning). See formatters::set_default_hsep() for further information.

...

ignored.

Details

When alt_counts_df is specified, column counts are calculated by applying the exact column subsetting expressions determined when applying column splitting to the main data (df) to alt_counts_df and counting the observations in each resulting subset.

In particular, this means that in the case of splitting based on cuts of the data, any dynamic cuts will have been calculated based on df and simply re-used for the count calculation.

Value

A TableTree or ElementaryTable object representing the table created by performing the tabulations declared in lyt to the data df.

Note

When overriding the column counts or totals care must be taken that, e.g., length() or nrow() are not called within tabulation functions, because those will NOT give the overridden counts. Writing/using tabulation functions which accept .N_col and .N_total or do not rely on column counts at all (even implicitly) is the only way to ensure overridden counts are fully respected.

Author(s)

Gabriel Becker

Examples

lyt <- basic_table() %>%
  split_cols_by("Species") %>%
  analyze("Sepal.Length", afun = function(x) {
    list(
      "mean (sd)" = rcell(c(mean(x), sd(x)), format = "xx.xx (xx.xx)"),
      "range" = diff(range(x))
    )
  })
lyt

tbl <- build_table(lyt, iris)
tbl

# analyze multiple variables
lyt2 <- basic_table() %>%
  split_cols_by("Species") %>%
  analyze(c("Sepal.Length", "Petal.Width"), afun = function(x) {
    list(
      "mean (sd)" = rcell(c(mean(x), sd(x)), format = "xx.xx (xx.xx)"),
      "range" = diff(range(x))
    )
  })

tbl2 <- build_table(lyt2, iris)
tbl2

# an example more relevant for clinical trials with column counts
lyt3 <- basic_table(show_colcounts = TRUE) %>%
  split_cols_by("ARM") %>%
  analyze("AGE", afun = function(x) {
    setNames(as.list(fivenum(x)), c(
      "minimum", "lower-hinge", "median",
      "upper-hinge", "maximum"
    ))
  })

tbl3 <- build_table(lyt3, DM)
tbl3

tbl4 <- build_table(lyt3, subset(DM, AGE > 40))
tbl4

# with column counts calculated based on different data
miniDM <- DM[sample(1:NROW(DM), 100), ]
tbl5 <- build_table(lyt3, DM, alt_counts_df = miniDM)
tbl5

tbl6 <- build_table(lyt3, DM, col_counts = 1:3)
tbl6