Heat demand prediction is an important part of increasing system efficiency within district heating. To achieve this efficiency, the energy provider companies need to estimate how much energy is re quired to satisfy the market demand. In this paper, we propose a method to investigate the application of online ma chine learning algorithm to achieve energy efficiency and optimization in District Heating (DH) systems by predicting the heat demand on the consumer side. To accomplish this, we are planning to use operational data from a Norwegian company (EffektivEnergi AS, Hamar) for a group of buildings that are connected to DH in other places.