In this presentation, I summarized our work on connecting prediction and prescription in energy management for residential microgrids, using stochastic optimization and machine learning to improve decision-making under uncertainty in systems with variable generation and storage.