Artificial Intelligence for Data Science in Theory and Practice

Preface
Acknowledgements
Contents
About the Editors
1 Machine Learning for Business Analytics: Case Studies and Open Research Problems
1.1 Introduction
1.2 ML Techniques in Business
1.2.1 Bayesian Classifiers
1.2.2 Decision Tree Algorithms
1.2.3 Logistic Regression
1.2.4 Support Vector Machines
1.2.5 Artificial Neural Networks
1.3 Machine Learning Applications in Business
1.3.1 Application-1: Development of a Repair Intelligence Platform Using Sensor Data, OEM Manuals, and Repair Invoices
1.3.2 Application-2: Machine Learning for Customer Support by Furniture Retailer (Faggella, 2018a)
1.3.3 Application-3: Business Intelligence Apps Built on Machine Learning (6 examples of AI in business intelligence applications, n.d.)
1.3.3.1 SAP Software Solutions: Machine Learning for Turning Databases into Useful Intel
1.3.3.2 Domo: Machine Learning for Business Dashboards
1.3.3.3 Apptus: Machine Learning in Sales Enablement
1.3.3.4 Avanade: Machine Learning for Business Insights
1.3.4 Application-4: Provider of Services for Smart Home Using Machine Learning (Faggella, 2018b)
1.3.5 Application-5: Machine Learning Powered Customer Sentiment Analysis by Microsoft (Faggella, 2018c)
1.4 Research Areas Related to Machine Learning for Business Data Analytics
1.4.1 Research Directions in Big Data Analytics with Hadoop (Lim et al., 2013)
1.4.2 Research Directions in Text Analytics
1.4.3 Research Directions in Network Analytics
1.4.3.1 Link Mining
1.4.3.2 Community Detection
1.4.3.3 Social Recommendation
1.4.4 Research Directions in Ethical Aspects of Data Collection for Business Analytics (Sivarajah et al., 2017)
1.4.4.1 Safeguarding the Personal Data of Customers
1.4.4.2 Customer Profiling
1.5 Conclusions
References
2 Past Achievements and Future Promises of Digital Transformation: A Literature Review
2.1 Introduction
2.2 Materials and Methods
2.2.1 Search Strategy
2.2.2 Inclusion and Exclusion Criteria
2.3 Results and Discussion
2.4 Conclusions
References
3 Algorithms for the Development of Deep Learning Models for Classification and Prediction of LearnerBehaviour in MOOCs
3.1 Introduction
3.1.1 Background
3.1.2 Interest
3.1.3 State of the Art
3.1.4 Our Contribution
3.1.5 Structure of the Document
3.2 Related Work
3.2.1 Outline
3.2.2 Overview of Learning Analytics
3.2.3 Importance of Student Success Predictive Models in MOOCs
3.2.4 Personalized Support and Interventions
3.2.5 Adaptive Content and Learner Pathways
3.2.6 Data Understanding
3.2.7 Common Metrics for Student Success in MOOCs
3.2.8 Inputs Used by Student Success Predictive Models
3.2.9 Activity-Based Models
3.2.10 Demographics-Based Models
3.2.11 Learning-Based Models
3.2.12 Discussion Forum and Text-Based Models
3.2.13 Cognitive Models
3.2.14 Social Models
3.2.15 Data Sources Providing Inputs/Features for Predictive Models
3.2.16 Features Engineering in Predictive Models
3.2.17 Relation Between Types of Model and the Outcome Predicted
3.2.18 Algorithms for Predictive Models and Metrics for Their Evaluation
3.2.19 Algorithms for Predictive Models
3.2.20 Metrics for Model Evaluation
3.2.21 Lessons Learned from Related Work
3.2.22 Approach
3.2.23 Context/Dataset
3.2.24 Method
3.2.24.1 Features Generation
3.2.25 Building the Model
3.3 Experimental Results and Discussion
3.4 Conclusion and Future Work
References
4 Rainfall Prediction Using Machine Learning Models: Literature Survey
4.1 Introduction
4.2 Methodology
4.3 Data Sets
4.4 Output Objectives
4.5 Input Features
4.6 Input Data Pre-processing
4.6.1 Data Imputation
4.6.2 Feature Selection/Reduction
4.6.3 Data Preparation for Classification
4.7 Machine Learning Techniques Used
4.8 Reporting of Results and Accuracy Measures
4.9 Discussion
4.10 Conclusions
Appendix 1: List of Abbreviations
Appendix 2: Summary Tables for References
References
5 Cognitive Computing, Emotional Intelligence, and Artificial Intelligence in Healthcare
5.1 Introduction
5.2 Methods
5.3 Results
5.4 Discussion
5.5 Conclusions
References
6 A Systematic Review on Application of Data Mining Techniques in Healthcare Analytics and Data-Driven Decisions
6.1 Introduction
6.1.1 Motivation and Scope
6.2 Methodology
6.2.1 Literature Search and Article Selection
6.2.2 Quality Assessment and Processing Steps
6.3 Results
6.3.1 Distribution of Papers Based on Year of Publication
6.3.2 Distribution of Papers Based on Publishing Journals
6.3.3 Healthcare Analytics Types Based on Literature Search Results
6.3.4 Application of Analytics in Healthcare
6.3.4.1 Cardiovascular Diseases
6.3.4.2 Diabetes
6.3.4.3 Cancer Diagnosis and Prediction Application
6.3.4.4 Healthcare Administration
6.3.4.5 Prognosis and Diagnosis
6.3.4.6 Pharmacovigilance
6.3.5 Theoretical Studies
6.4 Conclusion
References
7 Malaria Detection Using Machine Learning
7.1 Introduction
7.2 Review of Related Literature
7.3 Methodology
7.3.1 Datasets Used
7.3.2 Cell Segmentation Algorithm
7.3.3 Convolutional Neural Network Structure
7.3.4 Training of Base Model
7.3.5 Transfer Learning for Training of Specialized Model
7.3.6 Model Summary
7.4 Results
7.4.1 Base Model Training and Accuracy
7.5 Conclusions
References
8 Automatic Number Plate Recognition System for Oman
8.1 Introduction
8.2 Literature Review
8.2.1 Studies in Arabic-Speaking Countries
8.2.2 Studies in Other Countries
8.2.3 Summary of Prior Art Results
8.3 Materials and Methods
8.3.1 Overview
8.3.2 Number Plate Dataset
8.3.3 Number Plate Extraction Processing Stages
8.3.3.1 Input Vehicle Image
8.3.3.2 Bounding Box Creation
8.3.3.3 Closing Morphological Operations
8.3.3.4 Number Plate Rectangle Detection
8.3.3.5 Number Plate Cropping
8.3.4 Number Plate Recognition
8.3.4.1 Brightness and Contrast Adjustments
8.3.4.2 RGB to Grayscale Conversation
8.3.4.3 Image Binarization
8.3.5 Character Segmentation
8.3.6 Character Recognition Datasets
8.3.7 Character Recognition Algorithms
8.3.7.1 Overview of Character Recognition Subsystem
8.3.7.2 CNN Training and Evaluation
8.4 Results and Discussion
8.4.1 Outline of This Section
8.4.2 Number Plate Detection and Recognition Results
8.4.2.1 Image Processing Outcomes
8.4.2.2 Detection and Recognition Overall Accuracies
8.4.2.3 Factors that Affect Number Plate Detection and Recognition
8.4.2.4 Average Processing Time
8.4.3 Results of Using CNN Algorithm for Character Recognition
8.4.3.1 Training the CNN Model
8.4.3.2 Character Recognition Models' Performance
8.5 Conclusions
References
9 Real-Time Detection of First Stories in Twitter Using a FastText Model
9.1 Introduction
9.2 Related Work
9.3 Technical Background
9.3.1 Basic Algorithm for Real-Time First Story Detection
9.3.2 Term Frequency-Inverse Document Frequency
9.3.3 Modified Term Frequency-Inverse Document Frequency
9.3.4 Word Embeddings Using Word2vec
9.3.4.1 Overview
9.3.4.2 Word2vec Architecture
9.3.4.3 Continuous Bag-of-Words Model for Obtaining Vector Representation
9.3.4.4 Skip-Gram Model for Obtaining Vector Representations
9.3.4.5 Comparison Between CBOW and Skip-Gram
9.3.5 FastText
9.3.6 Storm
9.3.6.1 Overview of Storm
9.3.6.2 Components and Performance Metrics of Storm
9.4 Materials and Methods
9.4.1 Overview
9.4.2 FastText First Story Detection Implementation
9.4.2.1 Creating the Word Embedding Using FastText
9.4.3 Description of Dataset
9.5 Results
9.6 Conclusion and Future Work
Appendix
References
10 Using a Bi-Directional Long Short-Term Memory Model with Attention Mechanism Trained on MIDI Data for Generating Unique Music
10.1 Introduction
10.2 Algorithmic Composition
10.3 Related Work
10.4 The Problem of Music Generation
10.5 Dataset
10.6 Data Representation
10.6.1 Notes
10.6.2 Raw Audio
10.6.3 MIDI
10.6.4 Piano Rolls
10.7 Preparing the Data
10.7.1 Dataset Description
10.7.2 Data Encoding
10.7.3 Data Loading
10.8 Proposed Model
10.8.1 Long Short-Term Memory
10.8.2 Sequence-to-Sequence Generation Model
10.8.3 Attention Mechanism
10.8.4 Combining Seq2seq with Attention Mechanism
10.9 Model Construction and Workflows
10.10 Model Evaluation
10.11 Conclusion
10.12 Future Work
References
Index