Causal-Privacy Audit Workflow for Synthetic and Distilled Data in Dropout Support
A new workflow called CaP-Eval has been developed to audit synthetic and distilled student data for privacy and utility in institutional support decisions. The workflow evaluates data based on predictive accuracy, treatment-effect fidelity, robustness, and local training-record proximity. Results indicate that DPGNet and distilled data are more reliable for preserving financial-status treatment effects compared to adversarial and Gaussian Copula baselines, though distilled data retains a stronger proximity signal. AI