Skip to contents

Create wide format data following a given spec

Source: R/build_adsl.R, R/build_bds.R, R/build_occds.R, and 1 more
build_x.Rd

Extract and reshape data from a single data set according to the given specification as created by adam_spec_*(). In addition, an (updated) dictionary is returned along with the md5 checksum of the specified file.

Usage

build_adsl(spec)

build_bds(
  spec,
  dupl_ctrl = list(values_fn = NULL, arrange = NULL),
  names_ctrl = list(clean_fn = ~stringr::str_replace_all(.x, "[:punct:]|[:space:]", "_"),
    names_sep = "_"),
  rm = FALSE
)

build_occds(spec, values_fn_occds = NULL)

Arguments

spec

result of adam_spec_*()

dupl_ctrl

bds only. A list with two entries

  • values_fn function to handle duplicates in pivoting step. see details section for default.

  • arrange expression passed to arrange() optional sorting of data set prior to pivoting, e.g. in order to select the first/last value by date. defaults to NULL.

names_ctrl

bds only. A list with two entries handling cleaning and renaming of columns after pivoting

  • clean_fn defaults to stringr::str_replace_all(.x, '[:punct:]|[:space:]', '_').

  • names_sep defaults to '_'

rm

bds only. boolean. defaults to FALSE. if TRUE, a repeated measurement feature matrix with an additional .rmtime column is prepared. Only used, if is.null(spec$rm).

values_fn_occds

occds only. function that is used to summarize values in the pivoting step, if multiple rows per observation unit are present. if NULL, the maximum is used for numeric values and the last factor level is used for categorical values.

Value

A list with the following entries

  • data a tibble in wide format with one row per id

  • dict a tibble listing the distinct combinations of columns param, label, unit, time, column, source (if provided).

  • source a list passing the file slot from the given spec that the created data set is based upon along with the md5 checksum of this file if file was provided, NULL otherwise

  • flag_table build_adsl() only. flag table is passed from spec$flag_table slot

Details

Note that the output dictionary may differ from the dictionary created by adam_spec_*(), as multiple features may be derived from a single parameter at different time points.

values_fn is passed to pivot_wider(). The default is function(x) {ifelse(all(is.numeric(x)), mean(x, na.rm = TRUE), na.omit(x)[1])}

Authors

Maike Ahrens (ahrensmaike), Sebastian Voss (svoss09)