Название: Data-Enabled Analytics: DEA for Big Data
Автор: Joe Zhu, Vincent Charles
Издательство: Springer
Год: 2021
Страниц: 370
Язык: английский
Формат: pdf (true), epub
Размер: 36.1 MB

This book brings Data Envelopment Analysis (DEA) based techniques and Big Data together to explore the novel uses and potentials of DEA under Big Data. These areas are of widespread interest to researchers and practitioners alike. Considering the vast literature on DEA, one could say that DEA has been and continues to be, a widely used technique both in performance and productivity measurement, having covered a plethora of challenges and debates within the modelling framework.

Over the past four decades, DEA models have been applied in almost every major field of study. However, DEA has not been used to its fullest extent. As the inter- and intra-disciplinary research grows, DEA could be used in potentially many other ways; for instance, DEA could be viewed as a data mining tool for data-enabled analytics. One opportunity is brought by the existence of Big Data. Although Big Data has existed for a while now, gaining popularity among insight seekers, we are still in incipient stages when it comes to taking full advantage of its potential. Generally, researchers have either been interested in examining its origin or in developing and using Big Data technology.

As the amount of Big Data is growing every day in an exponential manner, so does its complexity; in this sense, various types of data are surfacing, whose study and examination could shed new light on phenomena of interest. A quick review of existing literature shows that big data is a new entrant within the DEA framework. Recently, there has been an increasing interest in bringing the two concepts together, with research studies aiming to integrate DEA and big data concepts within a single framework. But, more work is needed to fully explore the value of their intersection—it is time to view DEA in light of its potential usage in new fields or new usage within the existing fields, under the big data umbrella. It is time to view DEA models beyond their present scope and mine new insights for better data-driven decision-making.

In the chapter “Data Envelopment Analysis and Big dаta: A Systematic Literature Review with Repeated Bibliometric Analysis”, Vincent Charles, Tatiana Gherman, and Joe Zhu aim to identify the current avenues of research for studies integrating DEA with big data. The analysis performed shows that big data is a new entrant within the DEA literature, with the recent body of work in the field being indicative of an increasing interest in bringing the two concepts together under a single framework.

In the chapter “Acceleration of Large-Scale DEA Computations Using Random Forest Classification”, Anyu Yu, Yu Shi, and Joe Zhu propose a novel approach to accelerate DEA computations involving voluminous data. The proposed method uses random forest (RF) classification to predict and search for the best-practice decision-making units (DMUs) within the large-scale observations. The effectiveness of the proposed method is tested using numerical cases involving large-scale data. The authors find that the proposed DEA-RF method can decrease computation time significantly, while ensuring an acceptable level of accuracy.

In the chapter “The Estimation of Productive Efficiency Through Machine Learning Techniques: Efficiency Analysis Trees”, Juan Aparicio, Miriam Esteve, Jesus J. Rodriguez-Sala, and Jose L. Zofio revise the fundamentals of a new technique recently proposed in the literature for estimating production frontiers based on decision trees, called efficiency analysis trees (EAT), and extend it to the context of measuring productive efficiency under convexification, using the directional distance function. The authors further illustrate how the different methods work by resorting to two real datasets.