Synthetic population

Not to be confused with Synthetic population (biology).

Synthetic population is artificial population data that fits the distribution of people and their relevant characteristics living in a specified area as according to the demographics from census data.^[1] Synthetic populations are often a basis for microsimulation or also agent based models of population behavior.^[2] The latter can be used for simulation of disease transmission,^[3] traffic^[4] and similar.

Synthetic population are initial sets of agents with detailed demographic and socioeconomic attributes, which allow execution of agent-based microsimulation.^[5] Due to privacy reasons and data limitations and restrict observability of entire real population. Therefore, the population synthesis procedure is applied, which expands a small data sample of population by using auxiliary data, to generate a synthetic population as close as possible to the real population in its characteristics.

Examples of application

Chicago Social Interaction Model or chiSIM is an agent-based simulation of individuals and locations in Chicago along with their daily behavior. The population is modeled as a set of heterogeneous, interacting, adaptive agents. These agents are the population of all the residents of Chicago.^[6]

In 2023, World Data Lab created a synthetic population for New York using microdata and summary statistics.^[7] It was used to calculate the poverty levels among the neighborhoods for targeted social programs.