AMI data analytics: Customer charactrization by relief algorithm and supplementary tools

The present market environment is composed of different participants. Suppliers, demand aggregators, independent system operator Gencos, Transcos and so on, are the participants of this environment. Besides, in the literature of competitive electricity market, exchange of enough information among them is inevitable. Such information helps to better decision making for different aims such as cost decreasing, benefit maximizing, more accurate load forecasting, operation guides, planning and so on. This paper concentrates on a procedure helping better customer characterization by analyzing customers consumption data. For this goal, data-mining techniques in corporation with other algorithms are utilized. The result of this study is summarized in enough knowledge about customers consumption behavior classes and allocation of the best pattern for each individual class. Besides, in this study, for determination of the most affecting features, relief algorithm is utilized as a feature selection tool in comparison with PCA algorithm. Eventually, performance of the proposed algorithm is evaluated by using a real dataset related to Irish customers.