Potential Uses of Data Envelopment Analysis for Big Data and Machine Learning Integration in Evaluation

Simple item page
dc.contributor.authorSen, Anuradhaen
dc.contributor.committeechairArchibald, Thomas Greigen
dc.contributor.committeememberSunderman, Hannah Marieen
dc.contributor.committeememberFriedel, Curtis R.en
dc.contributor.committeememberMontrosse-Moorhead, Biancaen
dc.contributor.departmentAgricultural, Leadership, and Community Educationen
dc.date.accessioned2025-01-11T09:00:29Z
dc.date.available2025-01-11T09:00:29Z
dc.date.issued2025-01-10
dc.description.abstractData science and evaluation are two disciplines that could benefit enormously from successful integration. While data science can provide useful information in terms of statistics for any evaluation process, it as a subject can benefit from incorporating elements that investigate the value addition aspects addressed by evaluation and evaluative thinking. Evaluative thinking when incorporated with the analysis of big data sources can become a powerful tool in our hands that could be used in various sectors, for various purposes, to build equity and justice in society and make us better prepared to deal with exigencies. The new advancements in data science tools and approaches have attracted researchers from various disciplines, including evaluators. However social scientists and evaluators have been slow adopters of data science technologies compared to other fields. If evaluation practitioners and scholars do not pay attention, then the other fields will race ahead with different methods, models, tools, and approaches (Bamberger, Raftree, and Olazabal, 2016). The new and emerging data sources and analytical approaches can not only help the evaluation practitioners to solve problems of various dimensions, but also the integration of disciplines like data science with evaluation has the potential to benefit each other and as well as for the promotion of social good. Data Envelopment Analysis (DEA) is a nonparametric linear programming approach to measure the efficiency and productivity of decision-making units (DMUs), which can be used in both public and private sectors. This approach can enhance the decision-making process by utilizing new emerging data sources. DEA, after being introduced in the late 1970s, has been widely used in the management sector, and over the last few decades, it has gained popularity in various additional fields due to being a data-driven approach. As far back as 1986, DEA was the subject of an entire textit{New Directions for Evaluation} volume, but it has rarely been explored again in the field of evaluation since then. Especially given the increased availability and use of big data in evaluation, there is a need to revisit DEA and explore its potential applications and implications for evaluation. This study is threefold, where it first explores what the published literature reveals about the nature of DEA and its relationship to the core evaluation concepts of evaluative thinking, reasoning, and ethics, using a scoping review. Secondly, it demonstrates a proof of concept of how DEA could be applied and the implications for evaluative thinking and reasoning using big data. Lastly, this study seeks the perspectives of thought leaders on the application of data science technologies and DEA in evaluation.en
dc.description.abstractgeneralBig data is an important trend worldwide, and most of the innovations and discussions going on around big data are all happening outside evaluation, and this poses a double-sided risk. On one hand, if other fields like data science, management, banking, agriculture, chemistry, healthcare, etc. continue to innovate around big data without the understanding of evaluative thinking, reasoning, and ethics, then the applied inquiry that the researchers have developed and reflected upon within evaluation for decades, will be at risk due to an overly technical approach to data. On the other hand, the field of evaluation faces the problem of being left behind. For evaluation to remain relevant as a field, it must stay up to date with developments in adjacent disciplines. In other words, if evaluation practitioners and scholars do not pay attention, then the other fields will race ahead with different methods, models, tools, and approaches (Bamberger et al., 2016). It is a problem on two levels. One of the biggest challenges with big data is that they lack quality. It thus becomes imperative to make the best use of the data and to convey valid and compelling stories with these data, we must continue to explore different approaches to working with it. This paper explores an approach, Data Envelopment Analysis (DEA), to make sense of the data with potential contribution to decision-making. DEA is a nonparametric linear programming method for measuring the efficiency and productivity of decision-making units (DMUs) in both the public and private sectors. It has maintained popularity in various subject areas that use big data, like operations research, mathematics, environmental science, engineering, material science, etc. After being introduced in the late 1970s, it has been widely used in the management sector, and over the last few decades, has gained popularity in various additional fields due to being a data-driven approach. As far back as 1986, linear optimization, in particular DEA, was the subject of an entire textit{New Directions for Evaluation} volume, but it has rarely been explored again in the field of evaluation since then. Especially given the increased availability and use of big data in evaluation, there is a need to revisit DEA and explore its potential applications and implications for evaluation. As such, this study is divided into three parts, where the first part is a scoping review that explores what the published literature reveals about the nature of DEA and its relationship to the core evaluation concepts of evaluative thinking, reasoning, and ethics. Secondly, this study provides a demonstration of the application of DEA using big data in evaluation. And lastly, this study seeks the perspectives of thought leaders on the application of data science technologies and DEA in evaluation.en
dc.description.degreeDoctor of Philosophyen
dc.format.mediumETDen
dc.identifier.othervt_gsexam:42361en
dc.identifier.urihttps://hdl.handle.net/10919/124153
dc.language.isoenen
dc.publisherVirginia Techen
dc.rightsIn Copyrighten
dc.rights.urihttp://rightsstatements.org/vocab/InC/1.0/en
dc.subjectData Envelopment Analysisen
dc.subjectprogram evaluationen
dc.subjectbig dataen
dc.subjectmachine learningen
dc.subjectexpert interviewen
dc.subjectscoping reviewen
dc.subjectk-means clusteringen
dc.titlePotential Uses of Data Envelopment Analysis for Big Data and Machine Learning Integration in Evaluationen
dc.typeDissertationen
thesis.degree.disciplineAgricultural and Extension Educationen
thesis.degree.grantorVirginia Polytechnic Institute and State Universityen
thesis.degree.leveldoctoralen
thesis.degree.nameDoctor of Philosophyen

Files

Original bundle
Now showing 1 - 4 of 4
Name:
Sen_A_D_2025.pdf
Size:
3.73 MB
Format:
Adobe Portable Document Format
Download
Name:
Sen_A_D_2025_support_4.pdf
Size:
108.45 KB
Format:
Adobe Portable Document Format
Description:
Supporting documents
Download
Name:
Sen_A_D_2025_support_3.pdf
Size:
312.21 KB
Format:
Adobe Portable Document Format
Description:
Supporting documents
Download
Name:
Sen_A_D_2025_support_1.pdf
Size:
107.45 KB
Format:
Adobe Portable Document Format
Description:
Supporting documents
Download

Collections

Doctoral Dissertations