Naming Conventions

To maintain clarity across the multiple transformation steps (Splitting → Spy Injection → Augmentation), PAYN uses specific prefixes for all DataFrame artifacts.

Prefix	Meaning	Scientific Context
true_	Ground Truth	The original, unmodified data. Includes known positives and negatives (if available). Used for validation only. Example: true_train_data, true_test_data
spy_inf_	Spy Infused	Training data where a fraction of positives have been masked and injected into the unlabeled pool. Used to train the Spy Model. Example: spy_inf_train_data
augm_	Augmented	Data that has been processed by the Spy Model. Contains predicted probabilities and the "Reliable Negative" classification labels. Example: augm_validation_data