درخت تصمیم ایجاد شده توسط تکنیک Kohonen-CHAID تحت تاثیر عوامل ” تعطیلات، میانگین ارتفاع سقف ابر، کمینه دما و بیشینه دمای موثر”
ANALYSIS AND FORECASTING CONSUMPTION BEHAVIOR OF ELECTRICITY CUSTOMERS USING DATA MINING TECHNIQUES
(CASE STUDY: WEST AZARBAIJAN ELECTRICAL DISTRIBUTION COMPANY SUBSCRIBERS)
In this era, electrical energy is used more than other types of energies by human, to improve his welfare and to do routine activities. Due to the problem of saving electrical energy, it is vital to predict the amount of required load for having a reliable and stable distribution system. Consumers are one of major parts of electrical power supply chain. The aim of this study is to predict electricity consumption of consumers and analysis their consumption behavior affected by climatic factors. A proper and accurate prediction of energy consumption can prevent waste of financial resources due to increased operational costs. On the other hand, enormous amount of costumer consumption data requires modern IT tools such as Data Mining to analyze data. Data Mining is the process of discovering patterns and extracting knowledge from raw data.
In this study, we predict the consumption of consumers of West Azerbaijan Distribution Company. With respect to this and having 5595 costumer data in 12 two-month periods, we run prediction algorithms such as CHAID, C&R, Regression, Neural Networks on our data and evaluate the results with evaluation criteria like MAPE (Mean Absolute Percentage Error). Also, we study consumption behavior of consumers with clustering them using Kohonen algorithm. Finally we provide conclusion and suggestions for future studies.
Keywords: Electricity power consumption، Consumer، Data mining Prediction، Classification، Regression، Clustering