Data-Enabled Analytics: DEA for Big Data (International Series in Operations Research & Management Science, 312)

Preface
Contents
Data Envelopment Analysis and Big Data: A Systematic Literature Review with Bibliometric Analysis
	1 Introduction
	2 Methodology
	3 An Overview of the DEA and Big Data Literatures
		3.1 “DEA” Document Results (19,104 Documents)
		3.2 “Big Data” Document Results (98,501 Documents)
		3.3 A Brief Bibliometric Analysis of the DEA-Big Data Literature (67 Studies)
	4 A Systematic Literature Review of the DEA-Big Data Research Articles (35 Articles)
		4.1 A Brief Bibliometric Analysis of the DEA-Big Data Research Articles Composing the Final Sample (24 Articles)
		4.2 Thematic Analysis of the DEA-Big Data Research Articles (24 Articles)
			4.2.1 Purely Methodological Articles (3 Articles)
			4.2.2 Environmental Efficiency Evaluation (16 Articles)
			4.2.3 Other Types of Efficiency Evaluation (5 Articles)
	5 Discussion and Conclusion
	References
Acceleration of Large-Scale DEA Computations Using Random Forest Classification
	1 Introduction
	2 Methodology
	3 Numerical Examples
		3.1 Computer and Data Sample
		3.2 Numerical Case
		3.3 Comparisons of Different Machine Learning Methods
	4 Conclusion
	References
The Estimation of Productive Efficiency Through Machine Learning Techniques: Efficiency Analysis Trees
	1 Introduction
	2 Efficiency Analysis Trees
		2.1 The Efficiency Analysis Trees Algorithm
		2.2 Efficiency Analysis Trees (EAT) Efficiency Scores: Relating EAT with Free Disposal Hull (FDH) and (Convex) Data Envelopment Analysis (DEA)
	3 The Estimation of the Directional Distance Function Through EAT and CEAT
	4 Empirical Studies
		4.1 PISA Study
			4.1.1 Data
			4.1.2 Tree Growth
			4.1.3 Efficiency Results: Output Oriented EAT Vs. FDH and CEAT Vs. DEA
		4.2 Taiwanese Banks
			4.2.1 Data
			4.2.2 Tree Growth
			4.2.3 Efficiency Results: Directional EAT Vs. FDH and CEAT Vs. DEA
	5 The Algorithms EAT and Convex (EAT) in Python
	6 Conclusions and Future Works
	References
Hybrid Data Science and Reinforcement Learning in Data Envelopment Analysis
	1 Introduction
	2 Fundamentals
		2.1 Symbolic Regression
		2.2 Tree-Based Method
		2.3 Multivariate Adaptive Regression Spline (MARS)
		2.4 Reinforcement Learning
	3 Data Science with DEA
		3.1 Hybrid Data Science Framework
		3.2 Frontier Estimation with an Inefficiency Term
		3.3 Frontier Estimation with a Composite Error Term
	4 Reinforcement Learning with DEA
	5 Conclusion
	References
Aggregation of Outputs and Inputs for DEA Analysis of Hospital Efficiency: Economics, Operations Research and Data Science Perspectives
	1 Introduction
	2 Methodology
		2.1 Envelopment Estimators
		2.2 `Curse of Dimensionality'
		2.3 Aggregation of Inputs and Outputs Before DEA
			2.3.1 PCA-Based Aggregation
			2.3.2 Price-Based Aggregation
	3 Statistical Methods to Analyse the Efficiency Scores
		3.1 Kernel Density Estimation and Related Tests
		3.2 Analysis of Simple and Weighted Means
			3.2.1 Central Limit Theorems and Confidence Interval for Simple Means
			3.2.2 Central Limit Theorems and Confidence Intervals for Weighted Means
	4 Data and Variables
		4.1 Sample
		4.2 Variables
			4.2.1 Inpatient Output
			4.2.2 Outpatient Output
			4.2.3 Inputs
	5 A Comparison of Estimated Efficiency Scores Among Different Aggregation Approaches
	6 Concluding Remarks
	Appendix
		A Results for the Case of 3-Inputs and 1-Output Models
		B Results for the Case of 1-Input and 1-Output Models
	References
Parallel Processing and Large-Scale Datasets in Data Envelopment Analysis
	1 Introduction
	2 The Existing Methodologies
		2.1 The Standard Approach
		2.2 The Hierarchical Decomposition (HD) Procedure
		2.3 Build Hull (BH)
		2.4 Finding Reference Sets (FRS)
		2.5 The Enhanced Hierarchical Decomposition (EHD) Procedure
	3 Selecting an Appropriate Method
	4 Proposed Methods
		4.1 Introducing the Enhanced FRS (EFRS)
		4.2 Comparing EHD and EFRS
		4.3 Implementing EFRS into EHD
	5 Conclusion
	References
Network DEA and Big Data with an Application to the Coronavirus Pandemic
	1 Introduction
	2 Recent Work on DEA and Big Data
	3 Network DEA
	4 A Network Model of Covid
		4.1 The NDEA Model and Performance Measures
		4.2 Data and NDEA Performance Estimates
	5 Summary
	References
Hierarchical Data Envelopment Analysis for Classification of High-Dimensional Data
	1 Introduction
	2 Methodology
		2.1 Single-Level DEA
		2.2 Four-Level H-DEA
		2.3 Cluster Approach
			2.3.1 Multi-level Clustering Measurement
	3 The Results from the H-DEA Model
	4 Country Grouping Using Multi-Level K-Means Cluster Analysis
	5 Conclusions
	References
Dominance Network Analysis: Hybridizing Dea and Complex Networks for Data Analytics
	1 Introduction
	2 DNA: Methodology
	3 DNA: Application to COVID-19 Data
	4 Conclusions
	References
Value Extracting in Relative Performance Appraisal with Network DEA: An Application to U.S. Equity Mutual Funds
	1 Introduction
	2 Mutual Fund Performance Appraisal from Disbursement Management Perspective
	3 Network DEA Models for Technical Efficiency and Disbursement Efficiency Evaluation
		3.1 Technical Efficiency Evaluation (Fig. 4)
		3.2 Disbursement Efficiency Evaluation (Fig. 5)
		3.3 Disbursement Utilisation
		3.4 Effect of Random Variation on Disbursement Utilisation
		3.5 Extension to Input-Orientation Case
	4 Application to U.S. Equity Mutual Funds
		4.1 Comparison of MF Rankings Based on Technical Efficiency and Disbursement Efficiency Scores
		4.2 Practical Use of Disbursement Efficiency Score
	5 Extending to Assess Performance from Different Input Management Perspectives
	6 Robustness Check
		6.1 Comparison of the Results with Black Box Model
		6.2 Comparison of DEA-Based Performance and Risk-Adjusted Return Performance
	7 Concluding Remarks
	A.1 Appendices
		A.1.1 Appendix A (Table 10)
		A.1.2 Appendix B
			Network DEA Models for Performance Appraisal from Three Different Management Perspectives
	References
Measuring Chinese Bank Performance with Undesirable Outputs: A Slack-Based Two-Stage Network DEA Approach
	1 Introduction
	2 Literature Review
		2.1 Chinese Bank Efficiency in a Single Process Framework
		2.2 Bank Efficiency in a Two-Stage Process Framework
		2.3 Undesirable Outputs and Modelling in Two-Stage or Network Bank Efficiency
		2.4 The Contributions of this Paper
	3 The Slack-Based Two-Stage DEA Approach with Undesirable Outputs
		3.1 The Proposed UVSBM Model
		3.2 Properties of the UVSBM Model
		3.3 System and Stage Efficiencies of the Two-Stage System
		3.4 Variables and Data
		3.5 Overall Efficiency and Sub-stage Efficiencies for Chinese Listed Banks
		3.6 The Efficiency Changes for Chinese Listed Banks Over Time
		3.7 The Efficiency Comparisons for Different Types of Listed Banks
		3.8 The Frontier Projection of Two-Stage Processes
	4 Conclusions
	Appendix: Proof of Theorem
	References
Using Network DEA and Grey Prediction Model for Big Data Analysis: An Application in the Global Airline Efficiency
	1 Introduction
	2 Literature Review
	3 Research Design
		3.1 Definition of the Two-Stage Production Process for Airlines
		3.2 Checking the Model Validity
		3.3 Multiplicative Efficiency Aggregation with SOCP
		3.4 GM (1, N) Model – Grey Relational Model
		3.5 GM (1,1) – Grey Prediction Model
	4 Empirical Analysis
		4.1 Dynamic Efficiency Analysis
		4.2 Opening the Black Box of Overall Efficiency
		4.3 Performance Forecasting
	5 Concluding Remarks
	References
Index