nc (version 2025.3.24)

capture_melt_multiple: Capture and melt into multiple columns

Description

Match a regex to column names of a wide data frame (many columns/few rows), then melt/reshape the matching columns into multiple result columns in a taller/longer data table (fewer columns/more rows). Input should be a data frame with four or more regularly named columns of possibly different types to reshape, and output is a data table with at least two columns of reshaped data. For melting into a single result column, see capture_melt_single.

Usage

capture_melt_multiple(..., 
    fill = FALSE, na.rm = FALSE, 
    verbose = getOption("datatable.verbose"))

Value

Data table of reshaped/melted/tall/long data, with a new column for each unique value of the capture group named "column", and a new column for each other capture group.

Arguments

...

First argument must be a data frame to melt/reshape; column names of this data frame will be used as the subjects for regex matching. Other arguments (regex/conversion/engine) are passed to capture_first_vec along with nomatch.error=FALSE. The regex must define a group named "column" -- each unique value captured in this group becomes a column name for the reshaped data in the output. There must also be at least one other group, and the output will contain a column for each other group -- see examples.

fill

If TRUE, fill missing input reshape columns with runs of rows with missing values in the output reshape columns. Otherwise stop with an error (default).

na.rm

Remove missing values from melted data? (passed to melt.data.table)

verbose

Print verbose output messages? (passed to melt.data.table)

Author

Toby Hocking <[e-mail]> [aut, cre]

Examples

  data.table::setDTthreads(1)

  ## Example 1: melt iris columns to compare Sepal and Petal dims, as
  ## in cdata package, https://winvector.github.io/cdata/
  (iris.part.cols <- nc::capture_melt_multiple(
    iris,
    column=".*?",
    "[.]",
    dim=".*"))
  iris.part.cols[Sepal