Machine Learning in Educational Sciences: Approaches, Applications and Advances

✍ Scribed by Myint Swe Khine (editor)

• Publisher: Springer
• Year: 2024
• Tongue: English
• Leaves: 389
• Category: Library

✦ Synopsis

This comprehensive volume investigates the untapped potential of machine learning in educational settings. It examines the profound impact machine learning can have on reshaping educational research. Each chapter delves into specific applications and advancements, sheds light on theory-building, and multidisciplinary research, and identifies areas for further development. It encompasses various topics, such as machine-based learning in psychological assessment. It also highlights the power of machine learning in analyzing large-scale international assessment data and utilizing natural language processing for science education. With contributions from leading scholars in the field, this book provides a comprehensive, evidence-based framework for leveraging machine-learning approaches to enhance educational outcomes. The book offers valuable insights and recommendations that could help shape the future of educational sciences.

✦ Table of Contents

Preface
Contents
Editor and Contributors
Introduction
Exploring the Potential of Machine Learning in Educational Research
1 Introduction
2 Foundations of Machine Learning
3 Predicting Student Performance
4 Machine Learning in Assessment Processes
5 Machine Learning in Educational Research
6 Conclusion
References
Foundations of Machine Learning
An Introduction to Machine Learning for Educational Researchers
1 Defining Terms
2 Steps to Conduct ML
3 Evaluating Model Fit
4 Applied Example
5 Future Research
References
Machine Learning Applications in Higher Education Services: Perspectives of Student Academic Performance
1 Introduction
2 Background of the Research
3 Research Methodology
4 ML Application Trend in HE
5 Findings
6 Conclusion
References
Camelot: A Council of Machine Learning Strategies to Enhance Teaching
1 Introduction
2 Machine Learning Models: Theory
2.1 Supervised Model: Na¨ıve Bayes Classifier
2.2 Supervised Model: K Nearest Neighbors
2.3 Supervised Model: Support Vector Machines
2.4 Supervised Model: Logistic Regression
2.5 Supervised Model: Decision Trees
2.6 Supervised Model: Neural Networks
3 The Camelot Framework
3.1 Phase-1: Data Preprocessing
3.2 Phase-2: Model Training
3.3 Phase-3: Model Testing and Inference
4 Conclusion
Appendix A
References
Penalized Regression in Large-Scale Data Analysis
1 Introduction
2 Predictive Modeling
3 Basics of Penalized Regression
3.1 OLS Regression
3.2 Penalized Regression Using Convex Penalty Functions
4 Model Assessment
4.1 Cross-Validation
4.2 Akaike Information Criterion and Bayesian Information Criterion
4.3 Prediction Errors
4.4 Selection Counts
5 Extensions of Penalized Regression
5.1 Group Penalized Regression
5.2 Penalized Regression Using Concave Penalty Functions
5.3 GlmmLasso
5.4 Post-selection Inference (PSI)
6 Concluding Remarks
7 Coding Examples in R
7.1 LASSO and Enet Using glmnet()
7.2 MCP and Mnet Using grpreg()
References
Predicting Student Performance
Schools Students Performance with Artificial Intelligence Machine Learning: Features Taxonomy, Methods and Evaluation
1 Introduction
2 Taxonomy of Students’ Performance Factors
3 Related Work
4 Students Performance Features Exploration
5 Machine Learning Models for Students Performance Prediction
5.1 Support Vector Machine
5.2 Random Forest
6 Performance Analysis
6.1 Experimental Environment
6.2 Experiments
6.3 Experimental Results Analysis
6.4 Study of Data Without Previous Grades
7 Conclusions
References
Predicting Response Latencies on Test Questions Based on Features of the Questions
1 Introduction
2 Method
3 Data Source
3.1 Response Time
3.2 Cognitive Complexity of an Item
3.3 Linguistic Features
3.4 Question Design
4 Correlation Analysis
5 Traditional Regression Analysis
6 Machine Learning Methods
7 Results
8 Conclusions and Discussion
References
Predicting Student Attrition in University Courses
1 Addressing Student Drop-Out in Universities: Understanding the Underlying Causes and Implementing Multifaceted Solutions
2 How Can Machine Learning Help in this Dropout Problem for Both the Universities and the Students?
3 Categories of Factors that Influence Dropout
4 Models with Time-Invariant and Time-Variant Predictors
5 Models at Different Levels
6 Specialties of Educational Machine Learning Models
7 The Predictive Power of Models
8 Model Building Actors, Competences, and Exploitation of Results
9 Model Building
9.1 Embedded Machine Learning Models in an LMS
9.2 Levels of Indicators
9.3 Training-Prediction Schemes
9.4 Predictor Matrices and Target Vectors in the Different Schemes
9.5 Searching for the Optimal Model
10 Supervised Learning Algorithms
11 Performance Metrics and Their Use for Checking Model Bias and Variance
12 The Use of Performance Metrics in Model Training, Validation, and Testing Phase
13 Learning Curve of the Model
14 Conclusion
References
Improving Students’ Achievement Prediction in Blended Learning Environments with Integrated Machine Learning Methods
1 Introduction
2 Data Collection
3 Research Method and Results
3.1 Data Preprocessing and Model Debugging
3.2 Semantic Completeness Analysis
3.3 Semantic Matching Degree Analysis
3.4 Recognize Student Learning Pattern
3.5 Prediction Results Feedback to the Instruction Process
4 Conclusion
5 Limitations
Appendix A
References
Enhancing Predictive Performance in Identifying At-Risk Students: Integration of Topological Features, Node Embeddings in Machine Learning Models
1 Introduction
2 Literature Review
3 Methodology
3.1 Graph Representation
3.2 Topological Features
3.3 Node Embeddings
3.4 Feature Selection
3.5 Model Training and Evaluation
4 Results and Discussions
5 Conclusion
References
Machine Learning in Assessment Processes
Applying Topic Modeling to Understand Assessment Practices of U.S. College Instructors in Response to the COVID-19 Pandemic
1 Introduction
1.1 Assessment Amidst Emergency Remote Teaching During the Pandemic
2 Research Aims
3 Methods
3.1 Participants
3.2 Data Collection Procedure
3.3 Analytic Procedure
4 Results
4.1 Perceived Changes to Assessment Practices
4.2 Challenges Administering Assessments Online
5 Discussion
5.1 Implications and Recommendations for Practice
5.2 Limitations
5.3 Conclusion
Supplemental Materials
Understanding the Context of Teaching and Assessment
Effects of the Pandemic on Instructional Practices
Pre-pandemic Professional Development
References
Applying Machine Learning to Augment the Design and Assessment of Immersive Learning Experience
1 Introduction
2 Machine Learning Approaches in Educational Sciences Research
2.1 What is Machine Learning?
2.2 Application of Machine Learning in Educational Sciences
3 The Design and Assessment of Immersive Learning Experience
3.1 Immersive Learning experience—How Learning Occurs?
3.2 Augmenting the Design of Immersive Learning Experience with Machine Learning
3.3 Natural Language Processing and Conversational Artificial Intelligence
3.4 Assessment for Immersive Learning Experience
4 Discussion and Future Directions
5 Conclusion
References
Machine Learning in Educational Research
Machine Learning for Analyzing the Relationship Between Well-Being, Academic Performance with Large-Scale Assessment Data
1 Purposes
2 Theoretical Framework
3 Background of Well-Being in the Educational Context
3.1 Global Education Measurement
3.2 Well-Being and Academic Achievement
3.3 A Holistic Approach to Well-Being
3.4 Student Well-Being in the Context of Education 4.0
4 A Machine Learning Approach to Big Data
5 Method
5.1 Participants
5.2 Items of Concrete Scenarios as Independent Variables
5.3 Survey Items of Momentary Feeling
5.4 Plausible Value as Dependent Variable
5.5 Big Data Analytics
6 Results
6.1 Best Model: Boosting
6.2 Math Scores: Boosting Results
6.3 Science Scores: Boosting Results
6.4 Data Visualization: Median Smoothing
7 Discussion of Well-Being and Academic Performance
7.1 Peer Engagement and Academic Success
7.2 Negative Emotion
7.3 Limitations and Future Directions
8 Discussion of Machine Learning and Educational Research
Appendix A
References
Using Large Language Models to Probe Cognitive Constructs, Augment Data, and Design Instructional Materials
1 Motivation
2 Using AI to Assess Physics Problem-Solving
3 Research Questions
4 Method
4.1 ChatGPT as a Resource
4.2 A Textbook-Style Physics Problem
5 Findings
5.1 RQ1–ChatGPT’s Problem-Solving
5.2 RQ2–Prompting ChatGPT to Use Different Strategies
5.3 RQ3–Utilizing ChatGPT to Generate Instructional Materials
6 Discussion
Appendices
Appendix 1: Evaluation of the Solution
Appendix 2: Chat History on the Means-Ends Strategy
Appendix 3: Chat History on the “Working Backwards “ Strategy
Appendix 4a: Chat History on the “Plug and Chug"-Strategy
Appendix 4b: Chat History on the “Plug and Chug"-Strategy
References
Machine Learning Applications for Early and Real-Time Warning Systems in Education
1 Early Beginnings
2 Theoretical Underpinnings of the Scientific Framework for Predictive Analysis in Education
2.1 Model-Dependent Realism
2.2 Information Theory Perspective
3 Conceptual Bases for Developing Neural Networks.
3.1 Structure Neural Network (SNN)
3.2 Theoretical Framework of Academic Performance
4 A Systematic Procedure
5 Results
6 Discussion
References
Text Identification for Questions Generation According to Bloom's Taxonomy Using Natural Language Processing
1 Introduction
2 Automatic Generation of Concept Map
3 Natural Language Processing (NLP) for Automated Concept Map Generation
4 Experiment and Methodology
4.1 Framing of Questions Based on Bloom’s Taxonomy
4.2 Mapping of Course Learning Outcomes with Bloom’s Levels
4.3 Significance of Course Expert for Framing Questions
4.4 Block Diagram of Proposed System
5 Data Preprocessing, Training, and Validation
5.1 Data Preprocessing in MATLAB
5.2 Supporting Features
6 Construction of Bloom’s Taxonomy Classifier Using Long Short-Term Memory (LSTM)
6.1 Designing BiLSTM Architecture
6.2 Classification of Text in MATLAB Using Bidirectional LSTM (BiLSTM)
6.3 Training the Model Using BiLSTM (RNN)
7 Construction of Bloom’s Taxonomy Classifier Using Convolutional Neural Network (CNN)
7.1 Design of Network Architecture Using CNN
7.2 Training the Model Using CNN
8 Result and Discussion
8.1 Network Performance in BiLSTM Model
8.2 Network Performance in CNN Model
8.3 Testing the LSTM Model
8.4 Testing the CNN Model
9 Conclusion
References
Where Generative AI Fits Within and in Addition to Existing AI K12 Education Interactions: Industry and Research Perspectives
1 Introduction
2 RQ1: Analytical Overview of AI K12 Education Interactions
2.1 Learning with AI Applications: Student-Facing, Teacher-Facing
2.2 Learning About AI: AI Literacy for Students and Teachers
2.3 AIED Research Efforts that Drive Industry
2.4 AIED Research that Explores the Analogy Between AI and HI
3 RQ2: Analytical Overview of Generative AI K12 Education Interactions
3.1 Learning with Generative AI: Digital Content, Students, Teachers
3.2 Learning About Generative AI: AI Literacy Curriculum for Students and Teachers
3.3 Aied Research Efforts that Drives Industry: Using LLMS
3.4 AIED Research that Explores the Analogy Between AI and HI
4 Conclusion
5 Future Work and Limitations
References