Identify derived variables using recodeflow patterns

Identifies derived variables using recodeflow metadata patterns in variable_details. Derived variables are calculated from raw variables (e.g., BMI from height/weight) and should NOT be generated as mock data.

This function uses the recodeflow approach: derived variables are identified by metadata patterns in variable_details, NOT by role column flags.

Usage

identify_derived_vars(variables, variable_details)

Arguments

variables

data.frame. Variable-level metadata with column: variable

variable_details

data.frame. Required. Detail-level metadata with columns: variable, recStart, recEnd. Contains DerivedVar:: and Func:: patterns that identify derived variables.

Value

Character vector of derived variable names (may be empty if none found)

Details

Detection methods (recodeflow patterns only):

1. DerivedVar:: pattern: Variables with DerivedVar:: in recStart column

   • Example: recStart = "DerivedVar::[HWTGHTM, HWTGWTK]"

   • Indicates variable is derived from listed raw variables

2. Func:: pattern: Variables with Func:: in recEnd column

   • Example: recEnd = "Func::bmi_fun"

   • Indicates transformation function applied to derive variable

Why variable_details is required:

The recodeflow approach stores derivation logic in variable_details, not in variables. The role column (if present) is NOT used for derived variable detection. This ensures:

• Derivation logic is explicit (what variables, what function)

• Metadata is self-documenting

• Compatible with cchsflow and other recodeflow tools

Note: This function does NOT check role = "derived" even if present. Derived status is determined solely by DerivedVar:: and Func:: patterns.

See also

Other helpers: get_raw_var_dependencies()

Examples

if (FALSE) { # \dontrun{
# BMI derived from height and weight
variables <- data.frame(
  variable = c("height", "weight", "BMI_derived"),
  variableType = c("Continuous", "Continuous", "Continuous"),
  role = c("enabled", "enabled", "enabled"),  # NO "derived" flag needed
  stringsAsFactors = FALSE
)

variable_details <- data.frame(
  variable = c("height", "weight", "BMI_derived"),
  recStart = c("[1.4,2.1]", "[45,150]", "DerivedVar::[height, weight]"),
  recEnd = c("copy", "copy", "Func::bmi_fun"),
  stringsAsFactors = FALSE
)

identify_derived_vars(variables, variable_details)
# Returns: "BMI_derived"

# ADL derived from 5 ADL items
variable_details_adl <- data.frame(
  variable = c("ADL_01", "ADL_02", "ADL_03", "ADL_04", "ADL_05", "ADL_der"),
  recStart = c(rep("1", 5), "DerivedVar::[ADL_01, ADL_02, ADL_03, ADL_04, ADL_05]"),
  recEnd = c(rep("1", 5), "Func::adl_fun"),
  stringsAsFactors = FALSE
)

identify_derived_vars(variables, variable_details_adl)
# Returns: "ADL_der"
} # }