Machine Learning and Data Analytics for Solving Business Problems: Methods, Applications, and Case Studies

Description

This book presents advances in business computing and data analytics by discussing recent and innovative machine learning methods that have been designed to support decision-making processes. These methods form the theoretical foundations of intelligent management systems, which allows for companies to understand the market environment, to improve the analysis of customer needs, to propose creative personalization of contents, and to design more effective business strategies, products, and services. This book gives an overview of recent methods – such as blockchain, big data, artificial intelligence, and cloud computing – so readers can rapidly explore them and their applications to solve common business challenges. The book aims to empower readers to leverage and develop creative supervised and unsupervised methods to solve business decision-making problems. Preface Contents About the Editors 1 Predicting Salaries with Random-Forest Regression 1.1 Introduction 1.2 Related Work 1.3 Data Preparation and Random-Forest Regression Modelling 1.3.1 Data Source, Data Analysis and Feature Selection 1.3.1.1 Dependent Variable 1.3.1.2 Independent Variables 1.3.2 Outlier-Handling Strategy 1.3.3 Selection of a Machine-Learning Approach 1.3.3.1 Learning and Applying Random Forests 1.3.4 An Ensemble of Random-Forest Regression Models 1.4 Evaluation 1.4.1 Experimental Setup, Measure of Prediction Accuracy and Baseline 1.4.2 Experimental Results 1.4.3 Runtime 1.5 Discussion of the Results and Future Directions 1.5.1 Comparison to Related Work 1.5.2 Analysis of the Regression Trees and Random-Forest Results 1.5.3 Feature Importance 1.5.4 Usage of the Variable gender and Further Improvements 1.5.5 Assessment of Prediction Quality for Real-World Software Products and Questions of Deployment 1.6 Conclusion References 2 Data-Driven Analysis of Microfinance and Social Loans Before and During the COVID-19 Pandemic Using Exploratory Analysis and Decision Tree Classifiers 2.1 Introduction 2.2 Data Analytics Methodology for the Analysis of Microloans and Beneficiaries' Characteristics 2.2.1 Data Analytics Methodology 2.2.2 Data Collection and Preprocessing 2.3 Exploratory Analyses of Microloans Based on Credit Classes 2.3.1 Bivariate Analysis of Individual Credit Characteristics Based on Credit Classes 2.3.2 Bivariate Analysis of Business Credits Based on Credit Classes 2.4 Identification of the Main Changes in the Characteristics of Microloans Before and During the COVID-19 Pandemic Using Decision Tree Classifiers 2.4.1 Decision Tree: A Machine Learning Model to Solve Complex Classification Problems 2.4.2 Experiments Design and Empirical Results 2.5 Conclusion and Perspectives References 3 Identification of Credit Risks Using Cluster Analysis and Behavioural Scoring During the COVID-19 Pandemic 3.1 Introduction 3.2 Related Works on Behavioural Scoring and Machine Learning Usage for Building Credit Behavioural Scorecards 3.3 Identification of Credit Risk on the Basis of Cluster Analysis of Account Behaviours and Supervised Learning 3.4 Empirical Framework and Data Collection 3.4.1 Data Set Description 3.5 Empirical Results 3.6 Conclusion References 4 Improving Sales Prediction for Point-of-Sale Retail Using Machine Learning and Clustering 4.1 Introduction 4.2 Background 4.2.1 PoS Retail 4.2.1.1 Description of Physical Retailing 4.2.1.2 Fast-Moving Consumer Goods and ML 4.2.2 ML Approaches Applicable in Retail 4.2.2.1 Clustering Approaches 4.2.2.2 Classification Algorithms 4.2.2.3 Regression Algorithms 4.2.2.4 Deep Learning and Neural Network Methods 4.2.3 Case Presentation 4.3 Factors Affecting Product Sales in PoS Retail 4.3.1 Main Factors Affecting Sales and Location of Retail Stores 4.3.1.1 Demographic Factors for Clustering Stores 4.3.1.2 Performance Factors for Clustering Stores 4.4 Store Clusters 4.4.1 K-means Store Clusters 4.4.2 SOM (Kohonen) Store Clusters 4.5 Evaluation of Model Combinations for Sales Forecast 4.5.1 Cases of Model Combinations 4.5.2 Input Data Preparation 4.5.3 Evaluation of Model Combinations 4.6 Discussion of Results 4.7 Conclusions, Limitations and Future Work References 5 Telecom Customer Segmentation Using Deep Embedded Clustering Algorithm 5.1 Introduction 5.2 Related Work 5.3 Background 5.4 Methodology 5.5 Experimental Analysis 5.5.1 Dataset 5.5.2 Exploratory Analysis of the Dataset 5.5.3 Number of Clusters 5.5.4 Telecom Customer Segmentation 5.5.5 Analysis on Other Customer Segmentation Datasets 5.6 Conclusion and Future Scope References 6 Semantic Image Quality Assessment Using Conventional Neural Network for E-Commerce Catalogue Management 6.1 Introduction 6.2 Related Work 6.3 Proposed Approach 6.3.1 Transformation 6.3.1.1 Color Transformation 6.3.1.2 Scale Transformation 6.3.1.3 Contrast Transformation 6.3.1.4 Blur Transformation 6.3.2 Multi-CNN Architecture Design 6.3.3 Network Training 6.3.4 Deep Feature Dimension Reduction 6.3.5 Nonlinear Regression Prediction 6.4 Experimental Results and Analysis 6.4.1 Datasets 6.4.1.1 Evaluation Criteria 6.4.2 Experimental Result 6.5 Conclusion References 7 Contextual Recommender Systems in Business from Models to Experiments 7.1 Introduction 7.2 Recommender Systems: Notions and Concepts 7.2.1 Foundations of Recommender Systems 7.2.2 Classification of Recommender Systems 7.2.2.1 Content-Based Filtering 7.2.2.2 Collaborative Filtering 7.2.2.3 Hybrid Approaches 7.3 Context Awareness Recommender Systems 7.3.1 Definitions 7.3.2 Context-Aware Recommender Systems Approaches 7.3.3 Context-Aware Recommender Systems: Synthesis 7.4 Experimental Evaluation Process for CARS Systems 7.4.1 Datasets 7.4.2 Evaluation Methodology 7.4.2.1 Offline Evaluation 7.4.2.2 Online Evaluation 7.4.3 Evaluation Metrics 7.4.3.1 Prediction Accuracy Metrics 7.4.3.2 Top-N Metrics 7.4.3.3 Alternative Performance Metrics 7.4.4 Recommender Systems Platforms 7.4.5 Conventional Methods in Contextual Recommender Systems 7.5 Experimental Results: Case Study 7.5.1 Analyzing Parameter Sensitivity: Impact of the Number of Iterations 7.5.2 Results 7.6 CARS Systems on Business 7.7 Conclusion References 8 An Overview of Multi-View Methods for Text Clustering 8.1 Introduction 8.2 Overview of Multi-View Textual Data Clustering 8.2.1 Late Integration Based Methods 8.2.1.1 Ensemble Methods for Multi-View Text Clustering 8.2.1.2 Multi-View Clustering Based on Latent Models 8.2.2 Co-training Based Methods 8.2.2.1 Multi-View K-Means Based Methods 8.2.2.2 Self-Organizing Map Multi-View Clustering 8.2.2.3 Multi-View Spectral Clustering 8.2.3 Subspace Clustering Based Methods 8.2.3.1 Muti-View Subspace Clustering Based on Nonnegative Matrix Factorization 8.2.3.2 Multi-View Subspace Clustering Based on Shared Latent Representation 8.2.4 Summary of Multi-View Methods for Text Clustering 8.3 Experiments 8.3.1 Data Sets Description 8.3.2 Evaluation Measures 8.3.3 Experimental Results 8.4 Conclusion References 9 Real-Time K-Prototypes for Incremental Attribute Learning Using Feature Selection 9.1 Introduction 9.2 K-Prototypes and Streaming Data Preprocessing 9.2.1 Theoretical Concepts of K-Prototypes Algorithm 9.2.1.1 Algorithm 9.2.2 Streaming Data Preprocessing Techniques 9.2.2.1 Dimensionality Reduction for Data Streams 9.3 Proposed Feature Selection Incremental K-Prototypes 9.3.1 Feature Selection Incremental K-Prototypes Through Incremental Attribute Learning Context 9.3.1.1 Definition and Approach Presentation 9.3.2 Incremental K-Prototypes Through Incremental Attribute Learning Context 9.3.2.1 Algorithm 9.3.2.2 Merge Procedure 9.4 Experimentation 9.4.1 Framework 9.4.1.1 Real Data Sets Description 9.4.2 Evaluation Criteria 9.4.3 Results and Discussion 9.4.3.1 Sum of Squared Error Results 9.4.3.2 Davies-Bouldin Index Results 9.4.3.3 Run Time Results 9.5 Conclusion References 10 Applications of Industry 4.0 on Saudi Supply Chain Management: Technologies, Opportunities, and Challenges 10.1 Introduction 10.2 Industry 4.0 and Supply Chain Management (SCM) 10.2.1 Cyber-Physical Systems 10.2.2 Internet of Things 10.2.3 Blockchain 10.2.4 Artificial Intelligence 10.2.4.1 Machine Learning 10.2.4.2 Big Data Analytics 10.2.5 Cloud Technologies 10.3 Enablers of KSA SCM 4.0 10.4 Challenges of KSA SCM 4.0 10.5 Conclusion References Index