چكيده به لاتين
Abstract:
Energy is one of the major pillars of economic, industrial and scientific life in the world, and its absence leads to a crisis in the various dimensions of social, individual and industrial life. Hence, it is a victorious society that can postpone the crisis and use intelligent utilization of new technologies, proper consumption management and appropriate energy consumption patterns. The purpose of this study is as follows: (1) proposing a new data mining approach to identify and disclose knowledge with greater accuracy and depth, which was deployed on two datasets for different purposes to determine whether this approach has the efficiency and ability to discover knowledge with greater accuracy and depth or not. (2) identifying factors affecting energy consumption and efficiency in residential houses using a dataset which contains various information of residential buildings. (3) identifying and analyze the factors affecting household energy-related behaviors using a dataset, which contains comprehensive information of American households. The findings of this section suggest the usefulness and effectiveness of the proposed approach in exposing unknowns with greater depth and accuracy. In addition, due to the stated objectives, important factors were identified.
(4) An algorithm has been developed to improve the objects of clusters at micro and macro levels. This approach is consisting of two main steps, which in a sequential cycle lead to improvement within cluster and improvement between clusters, is presented. In the first step of this algorithm, cluster objects achieve better performance within its cluster by small changes. They actually approach to the center of their cluster. In the second step, the objects improve in such a way that they achieve better performance than their cluster performance and become more similar to another cluster objects. And then transfer to a better performance cluster. The advantage of the proposed algorithm is the dynamism of its thresholds, which can be changed depending on the objectives of the problem, budget, time, etc. The proposed algorithm was measured on the energy consumption of a dataset from American households. The results indicated that the proposed algorithm, in addition to improving households at micro and macro levels, which led to a reduction in energy consumption, also led to improvements in energy-related behaviors.
Keywords: Data mining, energy consumption, decision tree, associative rules, clustering improvement.
