Samar Turki

A Layered Decision-Aware Forecasting Architecture for Noisy Panel Environments

Machine learning forecasting systems in noisy, non-stationary, multi-entity environments are commonly optimized for pointwise predictive accuracy, although many downstream decisions depend on the relative ordering of entities rather than exact numerical forecasts. This creates a potential prediction–ranking–decision mi...