RESCAST-100K: A Comprehensive Dataset for Cross-Domain Residential Load and Indoor Temperature Forecasting
Researchers have introduced RESCAST-100K, a large-scale dataset designed to improve cross-domain forecasting for residential energy load and indoor temperature. The dataset, generated using EnergyPlus simulations for approximately 100,000 U.S. homes, includes detailed time-series data and building covariates. It also integrates five real-world datasets to facilitate sim-to-real evaluation, aiming to advance machine learning applications in home and grid-scale energy management. AI
IMPACT Enables more accurate cross-domain forecasting for residential energy management and grid response.