An Adaptive-Network-Based Fuzzy Inference System-Data Envelopment Analysis Algorithm for Optimization of Long-Term Electricity Consumption, Forecasting and Policy Analysis: The Case of Seven Industrialized Countries

Abstract

This article presents an adaptive-network-based fuzzy inference system (ANFIS)-data envelopment analysis (DEA) algorithm for improvement of long-term electricity consumption forecasting and analysis. Six models are proposed to forecast annual electricity demand. Six different membership functions and several linguistic variables are considered in building ANFIS. The proposed models consist of two input variables, namely, gross domestic product and population. All trained ANFIS are then compared with respect to mean absolute percentage error. To meet the best performance of the intelligent-based approaches, data are pre-processed (scaled) and finally our outputs are post-processed (returned to its original scale). DEA is used to optimize the electricity consumption as well as examine the behavior of electricity consumption. To show the applicability and superiority of the ANFIS-DEA algorithm, actual electricity consumption in the USA, Canada, Germany, United Kingdom (UK), Japan, France and Italy from 1980–2007 is considered. Electricity consumption is then forecasted up to 2015. The unique features of the ANFIS-DEA algorithm are: behavioral analysis and optimization in complex, non-linear and uncertain environments.