Parent
#91
What to build
Add three new fields to NumericImputationConfig that control n_nearest_features derivation for large MICE blocks. Update to_dict() / from_dict() to serialise them. Add unit tests confirming the round-trip preserves all three fields with their defaults.
Fields to add:
mice_n_nearest_features_min_cols: int = 10 — MICE block size below which n_nearest_features is left unset (all predictors used)mice_max_nearest_features: int = 20 — cap on the computed n_nearest_features valuemice_correlation_threshold: float = 0.1 — minimum |Pearson r| for a column to count as an informative predictor
Acceptance criteria
Blocked by
None — can start immediately
Parent
#91
What to build
Add three new fields to
NumericImputationConfigthat controln_nearest_featuresderivation for large MICE blocks. Updateto_dict()/from_dict()to serialise them. Add unit tests confirming the round-trip preserves all three fields with their defaults.Fields to add:
mice_n_nearest_features_min_cols: int = 10— MICE block size below whichn_nearest_featuresis left unset (all predictors used)mice_max_nearest_features: int = 20— cap on the computedn_nearest_featuresvaluemice_correlation_threshold: float = 0.1— minimum|Pearson r|for a column to count as an informative predictorAcceptance criteria
NumericImputationConfigwith the documented defaultsto_dict()includes all three fieldsfrom_dict()reconstructs all three fields; missing keys fall back to defaultsNumericImputationConfigis updated to document all three new parametersto_dict()/from_dict()round-trip preserves all three fields with correct defaultsfrom_dict()with an empty dict produces the correct defaults for all three fieldsBlocked by
None — can start immediately