Identification and Prediction of Consumption Behavior Using Decision Tree and Consumer Value Pyramid

What is important in the current situation it is important to pay attention to patterns of the water consumption behavior of subscribers and to identify subscribers and consumers with a higher position in the value pyramid in the field of water consumption management policy. Value pyramid is a tool that identifies valuable customers in terms of consumption, and so, has great application and importance in the field of water consumption management to identify high-consumption and low-consumption customers. Therefore, in the present study, in order to identify the patterns of consumption behavior of Shiraz Water and Wastewater Company subscribers based on their consumption value pyramid and to predict the subscribers with a higher position in the value pyramid, data mining techniques have been used. In the framework of the proposed method, first, the data of water subscribers' consumption including residential, commercial and industrial subscribers, public and administrative, religious and educational sites, military and non-governmental, for two consecutive years, were extracted from the database of Shiraz Water and Wastewater Company. After determining the optimal number of clusters using self-organizing map and Davis Bouldin index, clustering operation is performed using K-Means algorithm. It should be noted that the indicators and criteria for subscriber clustering include the type of use, location, consumption, history of unauthorized branching, number of disconnection notices and time of payment of bills which have been identified using the opinion of experts. Then, while calculating the consumption value of subscribers in each cluster and plotting the subscriber consumption value pyramid, the decision tree algorithm is used to predict and discover the behavioral patterns of subscribers. The results show that Shiraz Water and Wastewater Company subscribers are divided into six clusters in terms of consumption behavior patterns. While drawing the consumption value pyramid, these six clusters are classified into three classes: high consumption subscribers, medium consumption subscribers and low consumption subscribers. After implementing the decision tree, the accuracy of the tree was 78.92 that, according to the results of the decision tree, the subscribers of these three classes have 11 patterns of behavior that predict the type of consumption. Thus, according to the 11 behavioral patterns of the subscribers of Shiraz Water and Wastewater Company, the consumption of new subscribers can be predicted and its position in the value pyramid can be determined.