Data-Driven Targeting of Customers for Demand Response

Selecting customers for demand response (DR) programs is challenging, and existing methodologies are hard to scale and poor in performance. The existing methods are limited by lack of temporal consumption information at the individual customer level. We propose a scalable methodology for DR program targeting utilizing novel data available from individual-level smart meters. The approach relies on formulating the problem as a stochastic knapsack problem involving predicted customer responses. A novel and efficient approximation algorithm is developed so it can scale to problems involving millions of customers. The methodology is tested experimentally using real smart meter data in more than 58k residential households.