term · draft
Synthetic population
A made-up set of households and people for a place, built so that its totals and mixes match the census, letting a model hold everyone without holding anyone's real records.
A census publishes tables: how many households of each size live in each area, how many people of each age, how many work, how many own a car. It does not publish the people. A synthetic population fills that gap with invented households and residents whose counts, added up area by area, match the tables. No one in it is real, but the place as a whole looks right.
The usual recipe starts from a small anonymised sample of real records, such as the census microdata sample or a travel survey, and copies and reweights those records until they match the totals for each area. PopulationSim and simPop work this way. Some open populations come ready-made, such as the synthetic population of Canada and the Synthetic Population Catalyst for Great Britain.
A synthetic population is a starting point, not a model of behaviour. It says who lives where; something else has to say what they do. Travel models such as ActivitySim and MATSim, and epidemic models such as Covasim, take it as their input. It matches only the attributes it was fitted to, and it is as old as its census. Synthetic people describes the whole family.
- aka
- population synthesis, synthetic microdata, virtual population
discussed in
-
Models of a population at the level of its people: residents built from the census, run through their days, and given a voice by language models, with what the evidence says about each layer.
Tags: population synthesis, census, agent-based models, privacy