The Synthetic Population Catalyst (SPC) makes it easier for researchers to work with synthetic population data in England. It combines a variety of data sources and outputs a single file in protocol buffer format, describing the population in a given study area, with a particular focus on socio-economic characteristics and interactions between individuals. It is therefore well suited to create inputs for models studying the spreading of a pandemic or segregation (e.g.). The tool provides methods to export the outcome in diferent formats often use for researchers like CSV or JSON.
The input of the SPC tool is a list of the Middle Layer Super Output Area (MSOAs) where you want to create a spatially enriched sythetic population to feed other dynamic models. SPC includes a script to assist you with the proper list of the MSOAs by defining a Local Authority District area in England. Get started to download SPC data or run the tool in different MSOAs.
The history of this project is as follows:
- The Dynamic Model for Epidemics (DyME), originally written in R, then later converted to Python and OpenCL was first written: https://github.com/Urban-Analytics/RAMP-UA
- The "ecosystem of digital twins" branch heavily refactored the code to support running in different study areas and added a new seeding and commuting modelling: https://github.com/Urban-Analytics/RAMP-UA/tree/Ecotwins-withCommuting
- This separate repository was created to port the initialisation logic to Rust, following the above branch.
There are many contributors to the project through these different stages; the version control history can be seen on Github in the other repositories.
Synthetic data may propagate biases existing in the real data it is based on, introduce new ones, or remove useful outliers. See ONS ethical guidance for more details. SPC is based on a collection of different 'modelling modules', including some developed externally by other researchers. Each module is validated independently. Validation for newly created methods and links to previous projects can be found in the modelling methods.