Classification of New Active Consumers Performance According to Previous Events Using Decision Trees

With the growing concern regarding climate change, solutions such as distributed generation, namely renewable-based, increased in the energy system. However, their volatile behavior needed more flexibility from the demand side to balance – resorting to demand response programs. Active consumers play a critical role in this new paradigm. In this way, the uncertainty of their response to triggered events should be modeled. The authors developed a contextual consumer rate to properly select the participants in a demand response event according to their previous events in similar contexts. The innovation in the present paper lies in the classification of new active consumers with no prior experience. A decision tree method was then used to attribute a trustworthy rate. A sensitivity study on the number of leaf nodes used is explored. The results prove that the use of private information related to active consumer increase the performance of the algorithm.