AI-Powered Business Intelligence: Improving Forecasts and Decision Making with Machine Learning

Preface
    Who Should Read This Book
    Microsoft Power BI and Azure
    Learning Objectives
    Navigating This Book
    Conventions Used in This Book
    Using Code Examples
    O’Reilly Online Learning
    How to Contact Us
    Acknowledgments
1. Creating Business Value with AI
    How AI Is Changing the BI Landscape
    Common AI Use Cases for BI
        Automation and Ease of Use
        Better Forecasting and Predictions
        Leveraging Unstructured Data
    Getting an Intuition for AI and Machine Learning
    Mapping AI Use Case Ideas to Business Impact
    Summary
2. From BI to Decision Intelligence: Assessing Feasibility for AI Projects
    Putting Data First
    Assessing Data Readiness with the 4V Framework
        Combining 4Vs to Assess Data Readiness
    Choosing to Make or Buy AI Services
        AI as a Service
        Platform as a Service
        Infrastructure as a Service
        End-to-End Ownership
    Basic Architectures of AI Systems
        User Layer
        Data Layer
        Analysis Layer
    Ethical Considerations
    Creating a Prioritized Use Case Roadmap
        Mix Champions and Quick Wins
        Identify Common Data Sources
        Build a Compelling Vision
    Summary
3. Machine Learning Fundamentals
    The Supervised Machine Learning Process
        Step 1: Collect Historical Data
        Step 2: Identify Features and Labels
        Step 3: Split Your Data into Training and Test Sets
        Step 4: Use Algorithms to Find the Best Model
        Step 5: Evaluate the Final Model
        Step 6: Deploy
        Step 7: Perform Maintenance
    Popular Machine Learning Algorithms
        Linear Regression
        Decision Trees
        Ensemble Learning Methods
    Deep Learning
        Natural Language Processing
        Computer Vision
        Reinforcement Learning
    Machine Learning Model Evaluation
        Evaluating Regression Models
        Evaluating Classification Models
        Evaluating Multiclassification Models
    Common Pitfalls of Machine Learning
        Pitfall 1: Using Machine Learning When You Don’t Need It
        Pitfall 2: Being Too Greedy
        Pitfall 3: Building Overly Complex Models
        Pitfall 4: Not Stopping When You Have Enough Data
        Pitfall 5: Falling for the Curse of Dimensionality
        Pitfall 6: Ignoring Outliers
        Pitfall 7: Taking Cloud Infrastructure for Granted
    Summary
4. Prototyping
    What Is a Prototype, and Why Is It Important?
    Prototyping in Business Intelligence
    The AI Prototyping Toolkit for This Book
    Working with Microsoft Azure
        Sign Up for Microsoft Azure
        Create an Azure Machine Learning Studio Workspace
        Create an Azure Compute Resource
        Create Azure Blob Storage
    Working with Microsoft Power BI
    Summary
5. AI-Powered Descriptive Analytics
    Use Case: Querying Data with Natural Language
        Problem Statement
        Solution Overview
        Power BI Walk-Through
    Use Case: Summarizing Data with Natural Language
        Problem Statement
        Solution Overview
        Power BI Walk-Through
    Summary
6. AI-Powered Diagnostic Analytics
    Use Case: Automated Insights
        Problem Statement
        Solution Overview
        Power BI Walk-Through
    Summary
7. AI-Powered Predictive Analytics
    Prerequisites
    About the Dataset
    Use Case: Automating Classification Tasks
        Problem Statement
        Solution Overview
        Model Training with Microsoft Azure Walk-Through
        What Is an AutoML Job?
        Evaluating the AutoML Outputs
        Model Deployment with Microsoft Azure Walk-Through
        Getting Model Predictions with Python or R
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Use Case: Improving KPI Prediction
        Problem Statement
        Solution Overview
        Model Training with Microsoft Azure Walk-Through
        Model Deployment with Microsoft Azure Walk-Through
        Getting Model Predictions with Python or R
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Use Case: Automating Anomaly Detection
        Problem Statement
        Solution Overview
        Enabling AI Service on Microsoft Azure Walk-Through
        Getting Model Predictions with Python or R
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Summary
8. AI-Powered Prescriptive Analytics
    Use Case: Next Best Action Recommendation
        Problem Statement
        Solution Overview
        Setting Up the AI Service
        How Reinforcement Learning Works with the Personalizer Service
        Setting Up Azure Notebooks
        Simulating User Interactions
        Running the Simulation with Python
        Evaluate Model Performance in Azure Portal
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Summary
9. Leveraging Unstructured Data with AI
    Use Case: Getting Insights from Text Data
        Problem Statement
        Solution Overview
        Setting Up the AI Service
        Setting Up the Data Pipeline
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Use Case: Parsing Documents with AI
        Problem Statement
        Solution Overview
        Setting Up the AI Service
        Setting Up the Data Pipeline
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Use Case: Counting Objects in Images
        Problem Statement
        Solution Overview
        Setting Up the AI Service
        Setting Up the Data Pipeline
        Model Inference with Power BI Walk-Through
        Building the AI-Powered Dashboard in Power BI
    Summary
10. Bringing It All Together: Building an AI-Powered Customer Analytics Dashboard
    Problem Statement
    Solution Overview
    Preparing the Datasets
    Allocating a Compute Resource
    Building the ML Workflow
        Adding Sentiment Data to the Workflow
        Deploying the Workflow for Inference
    Building the AI-Powered Dashboard in Power BI
        Anomaly Detection
        Predictive Analytics
        AI-Powered Descriptive Analytics
        Unstructured Data
    Summary
11. Taking the Next Steps: From Prototype to Production
    Discovery Versus Delivery
    Success Criteria for AI Product Delivery
        People
        Processes
        Data
        Technology
    MLOps
    Get Started by Delivering Complete Increments
    Conclusion
Index
About the Author