Title: A neuro-fuzzy regression approach for estimation and optimisation of gasoline consumption

Authors: Ali Azadeh; S. Mohammad Hasan Manzour Alajdad; Tahereh Aliheidari Bioki

Addresses: School of Industrial Engineering and Center of Excellence for Intelligent Based Experimental Mechanic, College of Engineering, University of Tehran, P.O. Box 11365, Iran ' School of Industrial Engineering and Center of Excellence for Intelligent Based Experimental Mechanic, College of Engineering, University of Tehran, P.O. Box 11365, Iran ' Department of Economics, Science and Research Branch, Islamic Azad University, No. 4, Dr. Soltani Bldg, Daneshgah Blvd., Safaieh Sq, P.O. Box 89185/155, Yazd, Iran

Abstract: The purpose of the present study is to forecast the gasoline consumption of Iran. To this end, the economic indicators used in this paper are population, gross domestic production (GDP), natural income (NI), gasoline price, number of light vehicle, and production of gasoline in Iran. Various fuzzy regression (FR) models and also multiple train and transfer functions for estimating with artificial neural network (ANN), were used in this study and finally, linear regression for estimation of gasoline consumption was used. Five factors for comparing efficiency of fuzzy regression models were considered in the current case study. Furthermore, mean absolute percentage error (MAPE) for comparing efficiency of fuzzy regression, ANN and linear regression was selected. The FR, ANN, and linear regression models have been tuned for all their parameters according to the train data, following which the best coefficients and weights are identified. Three popular defuzzification methods for defuzzifying outputs are applied. For determining the rate of error of FR models estimations, the rate of defuzzified output of each model is compared with its actual rate consumption in test data and MAPE is calculated. The superiority and advantage of this study over previous studies is also presented.

Keywords: gasoline consumption; petrol consumption; fuzzy regression; fuzzy mathematical programming; artificial neural networks; ANNs; linear regression; fuzzy logic; Iran; optimisation.

DOI: 10.1504/IJSOM.2014.058844

International Journal of Services and Operations Management, 2014 Vol.17 No.2, pp.221 - 256

Published online: 17 Jun 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article