compare_native_vs_pmml_scores

Compare native T scoring vs JPMML scoring

Validates the T-native implementation of model scoring (e.g. random forest) against the reference JPMML evaluator. Returns a summary of differences.

Parameters

• df (DataFrame): The test data.

• model (Dict): The model dictionary with both native and PMML metadata.

Returns

A summary containing n_diffs, match (Bool), and n_rows.