Day 3: Modeling Concepts & Integration

Wednesday, February 4, 2026

Overview

Day 3 bridges machine learning and simulation modeling. We’ll explore different modeling paradigms and learn how to use ML predictions as inputs to agent-based models.

Learning Objectives

By the end of Day 3, you will be able to:

• Distinguish between System Dynamics and Agent-Based modeling approaches
• Understand when to use each modeling paradigm
• Generate predictions from ML models for simulation inputs
• Design the interface between ML and ABM components

Schedule

Time	Session	Lead	Materials
8:30-10:30	Modeling Concepts (SD vs ABM)	Ozawa	Slides
10:30-10:50	Break
10:50-12:30	Integration: ML to Simulation	Joint	Slides
12:30-1:30	Lunch
1:30-3:00	Supply Chain Simulation Foundations	Ozawa	Slides
3:00-3:20	Break
3:20-5:00	Hands-on: Introduction to ABM	Joint	Notebook

Materials

Slides

• Modeling Concepts - System Dynamics vs Agent-Based approaches
• ML-ABM Integration - Integrating Machine Learning and Agent-Based Models

Hands-on Exercise

• ML to ABM Integration Notebook

Key Concepts

System Dynamics vs Agent-Based Models

Aspect	System Dynamics	Agent-Based
Focus	Aggregate flows	Individual agents
Perspective	Top-down	Bottom-up
Data needs	Lower	Higher
Complexity	Simpler	More complex
Emergence	Not captured	Captured

Integration Workflow

ML Model (Demand Forecasting)
        ↓
    Predictions
        ↓
ABM Inputs (Agent behaviors, demand parameters)
        ↓
    Simulation
        ↓
   Policy Insights

The WISE Project Context

The WISE project aims to reduce wastage and shortages in Ethiopia’s pharmaceutical supply chain. The modeling approach:

1. ML Component: Forecast demand at different levels
2. ABM Component: Simulate supply chain dynamics
3. Integration: Use forecasts to parameterize agent behaviors
4. Policy: Test interventions in silico

Preparation

Before this session:

1. Complete Day 2 hands-on exercises
2. Have your trained ML model saved
3. Review glossary of simulation terms