An integrated artificial neural network-genetic algorithm clustering ensemble for performance assessment of decision making units

This study proposes a non-parametric efficiency frontier analysis method based on artificial neural network (ANN) and genetic algorithm clustering ensemble (GACE) for measuring efficiency as a complementary tool for the common techniques of the efficiency studies in the previous studies. The proposed ANN GA algorithm is able to find a stochastic frontier based on a set of input–output observational data and do not require explicit assumptions about the functional structure of the stochastic frontier. Furthermore, it uses a similar approach to econometric methods for calculating the efficiency scores. Moreover, the effect of the return to scale of decision making unit (DMU) on its efficiency is included and the unit used for the correction is selected based on its scale (under constant return to scale assumption). Also, in this algorithm, GA is used to cluster DMUs to increase A. Azadeh (B) Department of Industrial Engineering and Center of Excellence for Intelligent-Based Experimental Mechanics, College of Engineering, University of Tehran, P.O. Box 11365-4563, Tehran, Iran e-mail: [email protected]; [email protected] M. Saberi Institute for Digital Ecosystems & Business Intelligence, Curtin University of Technology, Perth, Australia e-mail: [email protected] M. Saberi Department of Industrial Engineering, Faculty of Engineering, University of Tafresh, Tafresh, Iran M. Anvari Department of Industrial Engineering, Iran University of Science and Technology, Tehran, Iran M. Mohamadi Department of Computer Engineering, Iran University of Science and Technology, Tehran, Iran DMUs’ homogeneousness. It should be noted that data envelopment analysis (DEA) is sensitive to the presence of the outliers and statistical noise. It is also not capable of performing prediction and forecasting. This is shown by two examples related to outlier situations. However, the proposed algorithm is capable of handling outliers and noise and DEA is used as a benchmark to show advantages of the proposed algorithm. Also, the proposed algorithm and conventional algorithm are compared in viewpoint of DEA through statistical t-test. The proposed approach is applied to a set of actual conventional power plants to show its applicability and superiority.