Skip to contents

Validate CDISC datasets against metadata-declared data types

Source: R/validate_cdisc_types.R
validate_cdisc_types.Rd

Checks each column of the supplied CDISC datasets against the expected storage class and maximum length declared in metadata_nca_variables. This is intentionally not a full CDISC conformance engine: it covers only what the metadata can assert (class and length), and never fails on metadata Type values it cannot interpret.

Usage

validate_cdisc_types(
  cdisc_data,
  metadata = metadata_nca_variables,
  check_length = TRUE,
  grouping_vars = character(0)
)

Arguments

cdisc_data

Named list of CDISC data frames keyed pp, adpp, and/or adnca (as returned by export_cdisc()). Entries with other names or that are not data frames are ignored.

metadata

Metadata data frame with Dataset, Variable, Type, and Length columns. Defaults to metadata_nca_variables.

check_length

Logical. Check character values against the declared Length? Defaults to TRUE.

grouping_vars

Character vector of explicitly selected grouping variables. Undeclared variables in this vector are allowed in ADPP and ADNCA when they contain supported atomic vectors, with a warning that metadata type/length checks are unavailable. PP remains strict. Known metadata rules always apply.

Value

A tidy findings data frame with columns Dataset, Variable, Check, Severity, Expected, Observed, N_Affected, and Message. Zero rows means the data conforms.

Details

Indexed CDISC variable families are resolved to their metadata template: a data column falls back from an exact name match to an index-agnostic match, so NCA1XRS/NCA2XRS match the NCAwXRS template and CRIT2/CRIT3FL match the CRIT1/CRIT1FL family. An exact metadata entry always takes precedence over the template.

Examples

cdisc_data <- list(
  adnca = data.frame(STUDYID = "S1", AVAL = 1.2, stringsAsFactors = FALSE)
)
validate_cdisc_types(cdisc_data)
#> [1] Dataset    Variable   Check      Severity   Expected   Observed   N_Affected
#> [8] Message   
#> <0 rows> (or 0-length row.names)