Tutorial: Agent-Based Modelling with Survey Data

This repository contains materials for a tutorial on Agent-Based Modelling (ABM), with applications to migration research.
It includes Julia code, Jupyter notebooks, R/Stata scripts, supporting survey data wrangling, and a presentation. To run survey ABM, LAMP data needs to be request to LAMP project administrators at https://mmp.research.brown.edu/.

---

Contents

1. Theoretical ABM

Folder: 1_theoretical_ABM/

• data/colinc/
  Household survey data on Colombian income.
• scenarios.jl and scenarios.ipynb
  Run all notebooks and simulate all scenarios.
• Other notebooks (e.g., agents.ipynb)
  Provide detailed explanations of specific components.

This section introduces the logic of a theoretical ABM and shows how to simulate scenarios step by step.

---

2. Survey ABM

Folder: 2_survey_ABM/

a. Data management and model preparation (2_survey_ABM/data_management_model/)

• run_scripts.R
  Loads lamp.rda and lamp_macro.rda and sources manage_lamp_ABM.R.
• manage_lamp_ABM.R
  Data wrangling of LAMP data to produce micro (lampABM.csv) and macro (macro.csv) datasets used in the ABM.
• model_ABM.do
  Stata script fitting a discrete-time event history model. Saves estimated migration probabilities to stat_model_lampABM_mlogit.csv.
• vars_interest/
  Selected variables from the LAMP survey data.

Outputs:

• lampABM.csv (micro-level data)
• macro.csv (macro-level data)
• stat_model_lampABM_mlogit.csv (estimated migration parameters)

b. ABM implementation (2_survey_ABM/)

• import.ipynb
  Creates functions to import micro and macro data.
• model.ipynb
  Defines behavioral rules and model parameters.
• setup.ipynb
  Setup functions for the ABM.
• run.ipynb
  Simulation functions, data transformations, and RMSE computation.
• scenarios Runs all notebooks; all scenarios are simulated.

Using survey-based ABM in migration research

This repository contains materials for a tutorial on Agent-Based Modelling (ABM), with applications to migration research.
It includes Julia code, Jupyter notebooks, R/Stata scripts, supporting survey data wrangling, and a presentation.