چكيده به لاتين
Energy is commonly known as the foundation for economic growth. A national energy policy is consequently of vital importance; it not only directs the development of a country but also influences the operational environment of different industries. Over the past few decades, energy consumption in the Islamic Republic of Iran has increased considerably due to the growing population and economic development. Natural gas is a major energy resource in the world and its consumption has had the quickest growth among fossil fuels in recent years. Iran is the world's fifth largest producer of natural gas; top on the list is Russia followed by US, EU, and Canada. Knowing the subscribers' consumption pattern can greatly help National Iranian Gas Company (NIGC) gain more awareness and intelligence in future decisions. This research aims to discover a consumption pattern in NIGC subscribers by designing a two-stage hybrid datamining model. In the first stage, k-means algorithm is used to cluster data. The clustering output and an ensemble learning method are then used in the second stage to classify subscribers according to their consumption behavior. This ensemble learning method makes use of majority voting technique. This study shows that ensemble learning method resulted from decision tree, nearest neighbor, and Naive Bayes, has higher precision and performance than each of the stated learners, and also support vector machine. Therefore, this method is applied in order to classify subscribers and discover their consumption pattern.
