Report Title: The use of machine learning techniques to estimate technical efficiency
Reported by: Professor Juan Aparicio (Miguel Hernandez University of Elche (UMH), Spain)
Beijing time: June 22, 2023 (Thursday) at 15:30 pm
Zoom: 835 4912 5671
Password: 230619
Report link: https://us06web.zoom.us/j/83549125671?pwd= N1ZCWWFNZm5POXlybE5kZlhKMjMwdz09
Report Summary:
Free Disposal Hull (FDH) and Data Envelopment Analysis (DEA) present the typical characteristics of a data-driven approach with the specific objective of determining technical efficiency and production frontiers in Engineering and Microeconomics. However, by construction, the frontier estimators generated by FDH and DEA suffer from overfitting problems; something that contrasts with currently accepted models in machine learning. In this regard, FDH and DEA can be seen as statistical descriptive tools that make up of a more complex approach, where the aim is to avoid overfitting in order to conclude something about the underlying Data Generating Process that is behind the generation of the observations in a production process. In this presentation, we show how Efficiency Analysis Trees (EAT), which is based on the adaptation of regression trees in Machine Learning, can be a possible solution to overcome the overfitting problem associated with FDH and DEA. Additionally, we show other alternative adaptations of well-known machine learning techniques with the objective of determining technical efficiency of a set of homogeneous production units. Furthermore, we illustrate how these machine learning-based techniques may be used as complement to the standard non-parametric methods through some empirical applications.
Reported by:
Juan Aparicio is a professor in the Department of Statistics, Mathematics and Information Technology of Miguel Hernandez University of Elche (UMH) in Spain, and also the director of the Operations research Center. He was co chairman of the Chairman of Efficiency and Productivity of Banco Santander (with Professor Knox Lovell). His research interests include efficiency and productivity analysis combined with machine learning and data science. In cooperation with Springer Press, he has independently or jointly edited several books, mainly focusing on the use of Data envelopment analysis for performance evaluation and benchmarking; And published about 150 scientific articles in different international journals. These journals include European Journal of Operational Research, OMEGA, Annals of Operations Research, International Journal of Production Economics, Journal of Optimization Theory and Applications, Journal of Productivity Analysis, Operational Research, Social Economic Planning Sciences, and Computers and Operations Research and Computers and Industrial Engineering. In particular, he recently published several adapted articles on different machine learning technologies to estimate the Production function and technical efficiency from the perspective of methodology. In addition, he also applied the new method to real databases in different departments such as education and banking. He has served as a keynote speaker at multiple conferences such as DEA International Conference in 2020. Finally, he is currently the deputy editor in chief of Omega, The International Journal of Management Science and Journal of Productivity Analysis.
(Undertaken by: Energy and Environmental Policy Research Center, Research and Academic Center)
