The procedure to generate a synthetic population has three main phases:
- Data preparation. It reads the input data (microdata, control totals, zone system, etc.) and prepares them for the optimization.
- Optimization. It calculates a set of weights of each one of the microdata records in an iterative approach to match the control totals. It uses iterative proportional update (IPU) at multiple levels and geographical resolutions.
- Allocation. It generates synthetic households, persons, dwellings, jobs and schools based on the weights at each specific geographical resolution. It uses Monte Carlo simulation with replacement.
Data preparation and allocation depend on the inputs and vary by implementation. Similarly, the geographical resolution of the optimization can vary by implementation.
The required inputs for the Munich implementation are: