Machine Learning for Oracle Database Professionals: Deploying Model-Driven Applications and Automation Pipelines

Table of Contents
About the Authors
About the Technical Reviewer
Acknowledgments
Introduction
	Readers and Audiences
Chapter 1: Introduction to Machine Learning
	Why Machine Learning?
	What Is Machine Learning?
		Supervised Learning
			Algorithms for Supervised Learning
		Unsupervised Learning
			Algorithms for Unsupervised Learning
		Semi-Supervised Learning
			Reinforcement Learning
			Self-Supervised Learning
	The Machine Learning Process
	Summary
Chapter 2: Oracle and Machine Learning
	Oracle Machine Learning for SQL (OML4SQL)
	Oracle and Other Programming Languages for Machine Learning
		R
		Python
		Java
	OCI Data Science
	Oracle Analytics Cloud
	AutoML
	Summary
Chapter 3: Oracle Machine Learning for SQL
	PL/SQL Packages for OML4SQL
		Privileges
		Data Dictionary Views
	Predictive Analytics
	Data Preparation and Transformations
		Understanding the Data
		Preparing the Data
	PL/SQL API for OML4SQL
		The Settings Table
		Model Management
		Model Evaluation
		Model Scoring and Deployment
		Partitioned Model
	Extensions to OML4SQL
		Oracle Data Miner and Oracle SQL Developer
		OML Notebooks
	Summary
Chapter 4: Oracle Autonomous Database for Machine Learning
	Oracle Cloud Infrastructure and Autonomous Database
		Oracle Cloud Infrastructure Services
		Sign-up and Access Oracle Cloud Infrastructure
	Oracle Autonomous Database Architecture and Components
		Oracle Autonomous Database Attributes
		Autonomous Database in Free Trier and Always Free
	Working with Oracle Autonomous Data Warehouse
		Provisioning Oracle Autonomous Data Warehouse
		Connect to Oracle Autonomous Data Warehouse
		Loading Data to Oracle Autonomous Data Warehouse
			Step 1: Upload a File from a Local Computer to Object Storage
			Step 2: Create a Credential
			Step 3: Load Data to a Table in Autonomous Data Warehouse
		Import Tables/Schema to Oracle Autonomous Database
	Oracle Machine Learning with ADW
		Accessing Oracle Machine Learning Through Oracle Autonomous Database
	Summary
Chapter 5: Running Oracle Machine Learning with Autonomous Database
	Oracle Machine Learning Collaborative Environment
		Starting with Oracle Machine Learning
		Sharing Workspaces with Other Users
		Creating a Machine Learning Notebook
		Specifying Interpreter Bindings and Connection Groups
	Running SQL Scripts and Statements
		Create and Execute SQL Scripts in a Notebook
		Run SQL Statements in a Notebook
	Work with Notebooks to Analyze and Visualize Data
	Summary
Chapter 6: Building Machine Learning Models with OML Notebooks
	Oracle Machine Learning Overview
		Supervised Learning and Unsupervised Learning
		Machine Learning Process Flow
	Oracle Machine Learning for SQL
		OML4SQL PL/SQL API and SQL Functions
		Data Preparation and Data Transformation
			Split Data
			Data Transformation
			Transformation Expressions
			Binning Transformations
		Model Creation
		Model Evaluation
			Model Application
			Result Comparison
		Model Scoring and Model Deployment
	An Example of Machine Learning Project
	Classification Prediction Example
		Data Preparation and Data Transformation
		Predicting Attribute Importance
		Model Creation
		Model Testing and Evaluation
		Model Application
	Summary
Chapter 7: Oracle Analytics Cloud
	Data Preparation
	Data Visualization and Narrate
	Machine Learning in Oracle Analytics Cloud
	Summary
Chapter 8: Delivery and Automation Pipeline in Machine Learning
	ML Development Challenges
		Classical Software Engineering vs. Machine Learning
		Model Drift
	ML Deployment Challenges
		ML Life Cycle
		Scaling Challenges
			Model Training
			Model Inference
			Input Data Processing
	Key Requirements
	Design Considerations and Solutions
		Automating Data Science Steps
		Automated ML Pipeline: MLOps
		Model Registry for Tracking
		Data Validation
		Pipeline Abstraction
		Automatic Machine Learning (AutoML)
		Model Monitoring
		Model Monitoring Implementation
	Scaling Solutions
		ML Accelerators for Large Scale Model Training and Inference
		Distributed Machine Learning for Model Training
		Model Inference Options
		Input Data Pipeline
	ML Tooling Ecosystem
		ML Platforms
		ML Development Tools
		ML Deployment Tools
	Summary
Chapter 9: ML Deployment Pipeline Using Oracle Machine Learning
	Mainstream ML Platforms
	Oracle Machine Learning Environment
	Data Extraction in Big Data Environment
	In-Cluster Parallel Data Processing
	Automated Data Preparation and Feature Engineering
		General Data Processing Automation
		Text Processing Automation
		AutoML
			Automated Model Selection
			Automated Feature Selection
			Automated Hyperparameter Tuning
	Scalable In-Database Model Training and Scoring
		In-Database Parallel Execution via Embedded Algorithms
			Task-Parallel Execution
			Data-Parallel Execution
			Degree of Parallelism
			Environments
			In-Database Parallel Execution with Partitioned Models
		In-Cluster Parallel Execution
	Model Management
	Saving Models Using R Datastores in Database
	Leveraging Open Source Packages
		TensorFlow Extended (TFX) for Data Validation
			Schema-Based Validation
			Training and Serving Skew Detection
			Drift Detection
		scikit-multiflow for Model Monitoring
		Kubeflow: Cloud-Native ML Pipeline Deployment
	Summary
Chapter 10: Building Reproducible ML Pipelines Using Oracle Machine Learning
	The Environment
		Setting up Oracle Machine Learning for R
		Verifying the Oracle Machine Learning for R Installation
			Verifying OML4R on the Server Side
			Verifying OML4R on the Client Side
		Setting up Open Source Components
	The Data
	Data Validation and Model Monitoring Implementation
		TensorFlow Data Validation (TFDV)
		Data Validation
		Model Monitoring
	Tracking and Reproducing ML Pipeline
		Data Version Control (DVC)
		Versioning Code, Data, and Model Files
		Demo with Actual ML Pipeline
			ML Pipeline Project with Git and DVC Initialization and Configuration
			Defining and Recording Dependencies with DVC
			Tracking and Reproducing ML Pipelines with DVC
			Sample Tracking Use Cases
			Reproducing ML Pipeline: An Example
				Step 1: Update the Pipeline and Track the Change Using Git
				Step 2: Reproduce the Pipeline Starting from the Evaluation Stage
			Separate Storage Locations for Code and Pipeline Artifacts
			Visualization of ML Pipeline
	OML4R Troubleshooting Tips
		Error When Connecting to Oracle Database (as oml_user)
			Solution
		Error Due to Missing Packages When Building Models
			Solution
		Error When Creating or Dropping R Scripts for Embedded R Execution
			Solution
	Summary
Index