Development of a Data Mining System for Subscriber Classification (Case Study: Electricity Distribution Company)

Currently, organizations and companies tend to provide customers with good and suitable services in accordance with their interests and behaviors. Thus, the better the customers are classified, the better the services provided will be. Data mining is an efficient process for helping companies discover patterns in the database and it is important to identify target customers in this process. In fact, customers are selected to provide new products and services. Customer classification is based on data mining techniques for customer identification. This study tends to classify customers using data mining algorithms such as decision tree CART, neural network and regression. The case study is customers of Electricity Distribution Company. Simulation results based on Clementine software show that population had the highest effect on the amount of power consumed in each of the six household, public, industrial, agricultural, road and commercial classes. This is consistent with the opinion of experts in the electric power industry, because higher number of subscribers of each class surely increases the amount of electricity consumed (not steadily). The second effective feature of power consumption in six classes is humidity, which in many classes has a relatively equivalent effect with the effect of temperature on power consumption