One-hour aggregated operational dataset for air-source heat pump heating setpoint decision support

The original monitoring data were recorded at a 3 min sampling interval, and every 20 consecutive samples were aggregated into a 60 min window. The dataset includes window-level variables derived from supply-water temperature, return-water temperature, indoor temperature, outdoor ambient temperature, heating temperature setpoint, electricity-meter increment, and temperature-difference-based features.

The dataset is intended for short-term heating energy-consumption forecasting, multi-task residual MLP–LSTM surrogate model training, and offline heating temperature setpoint screening under indoor thermal-state constraints.