Skip to content
Project

Behaviorally Grounded Synthetic Populations

Copyright

Marcel Bartumeu Ramentol

 Marcel Bartumeu Ramentol

Simulating how a city or country responds to a policy takes people, not just indicators: residents who share homes and budgets, belong to social networks, and follow daily routines tied to real places. Yet most places publish only aggregate statistics.

This project builds behaviorally grounded synthetic populations: complete, simulated populations with households, social ties and daily schedules, generated from data that exist almost anywhere. No personal records are used. Every value records what supports it, from official statistic to modeling assumption, so an assumption is never mistaken for a measured fact, and every release is frozen and reproducible.

The approach is built to travel: a new city or country comes in through its own local data, not new code. Andorra is the reference case, with a synthetic version of its entire population. Each population is tested against independent data it was never built on.

The populations are designed to plug into scenario and policy engines, so that projected futures become consequences for specific households, neighborhoods and groups of people.