SummaryTables Docstrings

Annotated(value, annotation; label = AutoNumbering())

Create an Annotated object which will be given a footnote annotation in the Table where it is used. If the label keyword is AutoNumbering(), annotations will be given number labels from 1 to N in the order of their appearance. If it is nothing, no label will be shown. Any other label will be used directly as the footnote label.

Each unique label must be paired with a unique annotation, but the same combination can exist multiple times in a single table.

Cell(value, style::CellStyle)
Cell(value; [bold, italic, underline, halign, valign, border_bottom, indent_pt, merge, mergegroup])

Construct a Cell with value value and CellStyle style, which can also be created implicitly with keyword arguments. For explanations of the styling options, refer to CellStyle. A cell with value nothing is displayed as an empty cell (styles might still apply). The type of value can be anything.

Some types with special behavior are:

• Multiline for content broken over multiple lines in a cell. This object may not be used nested in other values, only as the top-level value.
• Concat for stringing together multiple values without having to interpolate them into a String, which keeps their own special behaviors intact.
• Superscript and Subscript
• Annotated for a value with an optional superscript label and a footnote annotation.

CellStyle(;
    bold::Bool = false,
    italic::Bool = false,
    underline::Bool = false,
    halign::Symbol = :center,
    valign::Symbol = :top,
    indent_pt::Float64 = 0.0,
    border_bottom::Bool = false,
    merge::Bool = false,
    mergegroup::UInt8 = 0,
)

Create a CellStyle object which determines the visual appearance of Cells.

Keyword arguments:

• bold renders text bold if true.
• italic renders text italic if true.
• underline underlines text if true.
• halign determines the horizontal alignment within the cell, either :left, :center or :right.
• valign determines the vertical alignment within the cell, either :top, :center or :bottom.
• indent_pt adds left indentation in points to the cell text.
• border_bottom adds a bottom border to the cell if true.
• merge causes adjacent cells which are == equal to be rendered as a single merged cell.
• mergegroup is a number that can be used to differentiate between two otherwise equal adjacent groups of cells that should not be merged together.

Concat(args...)

Create a Concat object which can be used to concatenate the representations of multiple values in a single table cell while keeping the conversion semantics of each arg in args intact.

Example

Concat(
    "Some text and an ",
    Annotated("annotated", "Some annotation"),
    " value",
)
# will be rendered as "Some text and an annotated¹ value"

Specifies one variable to group over and an associated name for display.

Multiline(args...)

Create a Multiline object which renders each arg on a separate line. A Multiline value may only be used as the top-level value of a cell, so Cell(Multiline(...)) is allowed but Cell(Concat(Multiline(...), ...)) is not.

Replace(f, with)
Replace(f; with)

This postprocessor replaces all cell values for which f(value) === true with the value with. If with <: Function then the new value will be with(value), instead.

Examples

Replace(x -> x isa String, "A string was here")
Replace(x -> x isa String, uppercase)
Replace(x -> x isa Int && iseven(x), "An even Int was here")

Stores the index of the grouping variable under which the summaries defined in analyses should be run. An index of 0 means that one summary block is appended after all columns or rows, an index of 1 means on summary block after each group from the first grouping key of rows or columns, and so on.

Specifies one function to summarize the raw values of one group with, and an associated name for display.

function Table(cells;
    header = nothing,
    footer = nothing,
    round_digits = 3,
    round_mode = :auto,
    trailing_zeros = false,
    footnotes = [],
    postprocess = [],
    rowgaps = Pair{Int,Float64}[],
    colgaps = Pair{Int,Float64}[],
)

Create a Table which can be rendered in multiple formats, such as HTML or LaTeX.

Arguments

• cells::Vector{SpannedCell}: The list of SpannedCells that make up the table. No cells may overlap.

Keyword arguments

• header: The index of the last row of the header, nothing if no header is specified.
• footer: The index of the first row of the footer, nothing if no footer is specified.
• footnotes: A vector of objects printed as footnotes that are not derived from Annotated values and therefore don't get labels with counterparts inside the table.
• round_digits = 3: Float values will be rounded to this precision before printing.
• round_mode = :auto: How the float values are rounded, options are :auto, :digits or :sigdigits. If round_mode === nothing, no rounding will be applied and round_digits and trailing_zeros will have no effect.
• trailing_zeros = false: Controls if float values keep trailing zeros, for example 4.0 vs 4.
• postprocess = []: A list of post-processors which will be applied left to right to the table. A post-processor can either work element-wise or on the whole table object. See the postprocess_table and postprocess_cell functions for defining custom postprocessors.
• rowgaps = Pair{Int,Float64}[]: A list of pairs index => gap_pt. For each pair, a visual gap the size of gap_pt is added between the rows index and index+1.
• colgaps = Pair{Int,Float64}[]: A list of pairs index => gap_pt. For each pair, a visual gap the size of gap_pt is added between the columns index and index+1.

Round mode

Consider the numbers 0.006789, 23.4567, 456.789 or 12345.0.

Here is how these numbers are formatted with the different available rounding modes:

• :auto rounds to n significant digits but doesn't zero out additional digits before the comma unlike :sigdigits. For example, round_digits = 3 would result in 0.00679, 23.5, 457.0 or 12345.0. Numbers at orders of magnitude >= 6 or <= -5 are displayed in exponential notation as in Julia.
• :digits rounds to n digits after the comma and shows possibly multiple trailing zeros. For example, round_digits = 3 would result in 0.007, 23.457 or 456.789 or 12345.000. Numbers are never shown with exponential notation.
• :sigdigits rounds to n significant digits and zeros out additional digits before the comma unlike :auto. For example, round_digits = 3 would result in 0.00679, 23.5, 457.0 or 12300.0. Numbers at orders of magnitude >= 6 or <= -5 are displayed in exponential notation as in Julia.

ReplaceMissing(; with = Annotated("-", "- No value"; label = NoLabel()))

This postprocessor replaces all missing cell values with the value in with.

auto_round(number; target_digits)

Rounds a floating point number to a target number of digits that are not leading zeros. For example, with 3 target digits, desirable numbers would be 123.0, 12.3, 1.23, 0.123, 0.0123 etc. Numbers larger than the number of digits are only rounded to the next integer (compare with round(1234, sigdigits = 3) which rounds to 1230.0). Numbers are rounded to target_digits significant digits when the floored base 10 exponent is -5 and lower or 6 and higher, as these numbers print with e notation by default in Julia.

auto_round(        1234567, target_digits = 4) = 1.235e6
auto_round(       123456.7, target_digits = 4) = 123457.0
auto_round(       12345.67, target_digits = 4) = 12346.0
auto_round(       1234.567, target_digits = 4) = 1235.0
auto_round(       123.4567, target_digits = 4) = 123.5
auto_round(       12.34567, target_digits = 4) = 12.35
auto_round(       1.234567, target_digits = 4) = 1.235
auto_round(      0.1234567, target_digits = 4) = 0.1235
auto_round(     0.01234567, target_digits = 4) = 0.01235
auto_round(    0.001234567, target_digits = 4) = 0.001235
auto_round(   0.0001234567, target_digits = 4) = 0.0001235
auto_round(  0.00001234567, target_digits = 4) = 1.235e-5
auto_round( 0.000001234567, target_digits = 4) = 1.235e-6
auto_round(0.0000001234567, target_digits = 4) = 1.235e-7

listingtable(table, variable;
    rows = [],
    cols = [],
    summarize_rows = [],
    summarize_cols = [],
    variable_header = true,
    celltable_kws...
)

Create a listing table Table from table which displays raw values from column variable.

Arguments

• table: Data source which must be convertible to a DataFrames.DataFrame.
• variable: Determines which variable's raw values are shown. Can either be a Symbol such as :ColumnA, or alternatively a Pair where the second element is the display name, such as :ColumnA => "Column A".

Keyword arguments

• rows = []: Grouping structure along the rows. Should be a Vector where each element is a grouping variable, specified as a Symbol such as :Column1, or a Pair, where the first element is the symbol and the second a display name, such as :Column1 => "Column 1". Specifying multiple grouping variables creates nested groups, with the last variable changing the fastest.
• cols = []: Grouping structure along the columns. Follows the same structure as rows.
• summarize_rows = []: Specifies functions to summarize variable with along the rows. Should be a Vector, where each entry is one separate summary. Each summary can be given as a Function such as mean or maximum, in which case the display name is the function's name. Alternatively, a display name can be given using the pair syntax, such as mean => "Average". By default, one summary is computed over all groups. You can also pass Symbol => [...] where Symbol is a grouping column, to compute one summary for each level of that group.
• summarize_cols = []: Specifies functions to summarize variable with along the columns. Follows the same structure as summarize_rows.
• variable_header = true: Controls if the cell with the name of the summarized variable is shown.
• sort = true: Sort the input table before grouping. Pre-sort as desired and set to false when you want to maintain a specific group order or are using non-sortable objects as group keys.

All other keywords are forwarded to the Table constructor, refer to its docstring for details.

Example

using Statistics

tbl = [
    :Apples => [1, 2, 3, 4, 5, 6, 7, 8],
    :Batch => [1, 1, 1, 1, 2, 2, 2, 2],
    :Checked => [true, false, true, false, true, false, true, false],
    :Delivery => ['a', 'a', 'b', 'b', 'a', 'a', 'b', 'b'],
]

listingtable(
    tbl,
    :Apples => "Number of apples",
    rows = [:Batch, :Checked => "Checked for spots"],
    cols = [:Delivery],
    summarize_cols = [sum => "overall"],
    summarize_rows = :Batch => [mean => "average", sum]
)

postprocess_cell

Overload postprocess_cell(c::Cell, postprocessor::YourPostProcessor) to enable using YourPostProcessor as a cell postprocessor by passing it to the postprocess keyword argument of Table.

The function must always return a Cell. It will be applied on every cell of the table that is being postprocessed, all other table attributes will be left unmodified.

Use postprocess_table instead if you need to modify table attributes during postprocessing.

postprocess_table

Overload postprocess_table(t::Table, postprocessor::YourPostProcessor) to enable using YourPostProcessor as a table postprocessor by passing it to the postprocess keyword argument of Table.

The function must always return a Table.

Use postprocess_cell instead if you do not need to modify table attributes during postprocessing but only individual cells.

summarytable(table, variable;
    rows = [],
    cols = [],
    summary = [],
    variable_header = true,
    celltable_kws...
)

Create a summary table Table from table, which summarizes values from column variable.

Arguments

• table: Data source which must be convertible to a DataFrames.DataFrame.
• variable: Determines which variable from table is summarized. Can either be a Symbol such as :ColumnA, or alternatively a Pair where the second element is the display name, such as :ColumnA => "Column A".

Keyword arguments

• rows = []: Grouping structure along the rows. Should be a Vector where each element is a grouping variable, specified as a Symbol such as :Column1, or a Pair, where the first element is the symbol and the second a display name, such as :Column1 => "Column 1". Specifying multiple grouping variables creates nested groups, with the last variable changing the fastest.
• cols = []: Grouping structure along the columns. Follows the same structure as rows.
• summary = []: Specifies functions to summarize variable with. Should be a Vector, where each entry is one separate summary. Each summary can be given as a Function such as mean or maximum, in which case the display name is the function's name. Alternatively, a display name can be given using the pair syntax, such as mean => "Average". By default, one summary is computed over all groups. You can also pass Symbol => [...] where Symbol is a grouping column, to compute one summary for each level of that group.
• variable_header = true: Controls if the cell with the name of the summarized variable is shown.
• sort = true: Sort the input table before grouping. Pre-sort as desired and set to false when you want to maintain a specific group order or are using non-sortable objects as group keys.

All other keywords are forwarded to the Table constructor, refer to its docstring for details.

Example

using Statistics

tbl = [
    :Apples => [1, 2, 3, 4, 5, 6, 7, 8],
    :Batch => [1, 1, 1, 1, 2, 2, 2, 2],
    :Delivery => ['a', 'a', 'b', 'b', 'a', 'a', 'b', 'b'],
]

summarytable(
    tbl,
    :Apples => "Number of apples",
    rows = [:Batch],
    cols = [:Delivery],
    summary = [length => "N", mean => "average", sum]
)

table_one(table, analyses; keywords...)

Construct a "Table 1" which summarises the patient baseline characteristics from the provided table dataset. This table is commonly used in biomedical research papers.

It can handle both continuous and categorical columns in table and summary statistics and hypothesis testing are able to be customised by the user. Tables can be stratified by one, or more, variables using the groupby keyword.

Keywords

• groupby: Which columns to stratify the dataset with, as a Vector{Symbol}.
• nonnormal: A vector of column names where hypothesis tests for the :nonnormal type are chosen.
• minmax: A vector of column names where hypothesis tests for the :minmax type are chosen.
• tests: a NamedTuple of hypothesis test types to use for categorical, nonnormal, minmax, and normal variables.
• combine: an object from MultipleTesting to use when combining p-values.
• show_overall: display the "Overall" column summary. Default is true.
• show_n: Display the number of rows for each group key next to its label.
• show_pvalues: display the P-Value column. Default is false.
• show_testnames: display the Test column. Default is false.
• show_confints: display the CI column. Default is false.
• sort: Sort the input table before grouping. Default is true. Pre-sort as desired and set to false when you want to maintain a specific group order or are using non-sortable objects as group keys.

All other keywords are forwarded to the Table constructor, refer to its docstring for details.

to_docx(ct::Table)

Creates a WriteDocx.Table node for Table ct which can be inserted into a WriteDocx document.

Note that you also have to insert the styles from SummaryTables.docx_styles() to the style section of the WriteDocx document.