Smart meter data is analyzed in residential buildings using machine learning algo-rithms. Individual consumers (heat pumps, water heaters, e-mobiles, charging stations, ...) and producers (photovoltaic systems) are identified. Valuable infor-mation about the disaggregated consumers is determined, such as their maxi-mum power consumption, the switch-on and switch-off times and the energy con-sumption per day. This data can be used to generate load profile forecasts with the aid of temperature, solar radiation and weather forecasts.