Preflight readiness check for ODA / CTA analysis

Validates a predictor frame, class vector, and optional weight vector before fitting. Returns a structured report. Does not modify inputs.

Usage

oda_readiness_check(
  X,
  y,
  w = NULL,
  miss_codes = NULL,
  binary_only = FALSE,
  min_class_n = 5L
)

Arguments

X

Data frame of predictors.

y

Integer class/group vector.

w

Optional numeric weight vector.

miss_codes

Numeric vector of missing-code values (default NULL).

binary_only

Logical; flag > 2 classes as an issue (default FALSE).

min_class_n

Minimum observations per class; flags if any class is below this threshold (default 5L).

Value

Named list with: ok (logical, TRUE if no issues), issues (character vector), warnings (character vector, non-fatal), n_obs (integer), group_report (from oda_validate_group()), weight_report (from oda_validate_weights()), attr_types (from oda_infer_attr_types()), constant_attrs (character vector of constant columns).

Details

Flags:

• Missing class/group variable.

• Non-binary group when binary_only = TRUE.

• Non-numeric weights, wrong-length weights, NA/Inf/zero weights.

• Missing-code patterns in predictors (if miss_codes supplied).

• Constant attributes (zero variance after miss-code removal).

• Insufficient class counts (< min_class_n).

• Attribute-type uncertainty (logical/factor columns).

See also

oda_validate_group, oda_validate_weights, oda_infer_attr_types, oda_clean_missing_codes