Free book: Bayesian Networks & BayesiaLab: A Practical Introduction for Researchers

Table of Contents

1. Introduction

• All Roads Lead to Bayesian Networks
• A Map of Analytic Modeling

2. Bayesian Network Theory

• A Non-Causal Bayesian Network Example
• A Causal Network Example
• A Dynamic Bayesian Network Example
• Representation of the Joint Probability Distribution
• Evidential Reasoning
• Causal Reasoning
• Learning Bayesian Network Parameters
• Learning Bayesian Network Structure
• Causal Discovery

3. BayesiaLab

• BayesiaLab’s Methods, Features, and Functions
• Knowledge Modeling
• Discrete, Nonlinear and Nonparametric Modeling
• Missing Values Processing
• Parameter Estimation
• Bayesian Updating
• Machine Learning
• Inference: Diagnosis, Prediction, and Simulation
• Model Utilization
• Knowledge Communication

4. Knowledge Modeling & Reasoning

• Background & Motivation
• Example: Where is My Bag?
• Knowledge Modeling for
  Problem #1
• Evidential Reasoning for
  Problem #1
• Knowledge Modeling for
  Problem #2
• Evidential Reasoning for
  Problem #2

5. Bayesian Networks and Data

• Example: House Prices in Ames, Iowa
• Data Import Wizard
• Discretization
• Graph Panel
• Information-Theoretic Concepts
• Parameter Estimation
• Naive Bayes Network

6. Supervised Learning

• Example: Tumor Classification
• Data Import Wizard
• Discretization Intervals
• Supervised Learning
• Model 1: Markov Blanket
• Model 1: Performance Analysis
• K-Folds Cross-Validation
• Model 2: Augmented Markov Blanket
• Cross-Validation
• Structural Coefficient
• Model Inference
• Interactive Inference
• Adaptive Questionnaire
• WebSimulator
• Target Interpretation Tree
• Mapping

• Example: Stock Market
• Data Import
• Data Discretization
• Unsupervised Learning
• Network Analysis
• Inference with Hard Evidence
• Inference with Probabilistic and Numerical Evidence
• Conflicting Evidence

8. Probabilistic Structural Equation Models

• Example: Consumer Survey
• Workflow Overview
• Data Import
• Step 1: Unsupervised Learning
• Step 2: Variable Clustering
• Step 3: Multiple Clustering
• Step 4: Completing the Probabilistic Structural Equation Model
• Key Drivers Analysis
• Multi-Quadrant Analysis
• Product Optimization

9. Missing Values Processing

• Types of Missingness
• Missing Completely at Random
• Missing at Random
• Missing Not at Random
• Filtered Values
• Missing Values Processing in BayesiaLab

10. Causal Identification & Estimation

• Motivation: Causality for Policy Assessment and Impact Analysis
• Sources of Causal Information
• Causal Inference by Experiment
• Causal Inference from Observational Data and Theory
• Identification and Estimation Process
• Causal Identification
• Computing the Effect Size
• Theoretical Background
• Potential Outcomes Framework
• Causal Identification
• Ignorability
• Example: Simpson’s Paradox
• Methods for Identification and Estimation
• Workflow #1: Identification and Estimation with a Directed Acyclic Graph
• Indirect Connection
• Common Parent
• Common Child (Collider)
• Creating a CDAG Representing Simpson’s Paradox
• Graphical Identification Criteria
• Adjustment Criterion and Identification
• Workflow #2: Effect Estimation with Bayesian Networks
• Creating a Causal Bayesian Network
• Path Analysis
• Pearl’s Graph Surgery
• Introduction to Matching
• Jouffe’s Likelihood Matching
• Direct Effects Analysis