Projects / Programmes
Synthetic Population Generation as the Basis for Activity-based/Agent-based Micro-simulation Transport Models
| Code |
Science |
Field |
Subfield |
| 2.19.01 |
Engineering sciences and technologies |
Traffic systems |
Traffic technics and technology |
| Code |
Science |
Field |
| T260 |
Technological sciences |
Physical planning |
| Code |
Science |
Field |
| 2.01 |
Engineering and Technology |
Civil engineering |
travel habits, travel behavior, agent-based microsimulation traffic models, synthesized population, IPF - Iterative Proportional Fitting Methods, SRM - Synthetic Reconstruction Methods, COT - Combinatorial Optimization Techniques
Researchers (19)
Organisations (3)
Abstract
Objective of research is of study is fundamental knowledge and understanding of travel/traffic behavior/patterns.
Estimations and forecasts of travel demand and behavior are usually handled by a standard methodological approaches, commonly referred as four-step modelling approach. Mostly they are chosen because of their convenient mathematical calculus, relative simplicity and easy understanding. However, such predictions in the past gave many substantial errors in model forecasts. Increased concern about phenomena such as congestion, emissions and changing land-use patterns forced governments to tackle them. Lately more importance is given to travel demand management, objective of what is to i) alter travel behavior without necessarily embarking on large-scale infrastructure expansion projects, ii) better use of available transport resources and iii) avoid the negative consequences of continued unrestrained growth in private mobility. Consequently there is growing need for the implementation of »agent-based« micro-simulation transport models, which i) are much more realistic (comparing to four-step models), ii) understand decision making process of individuals and iii) are responsive to a wider range of transport policy measures.
The initial step of such micro-simulation models is the definition of agents and relationships between them. Adequate data source is national census. However, census data must be prepared in order to be useful as input for micro-simulation. Complete census data is rarely available. More often, just small sub-sample can be obtained. Forecasts and predictions made on the basis of such sub-samples can lead to wrong conclusions. With generating of synthetic population it is possible to avoid such issues.
Main objective is to generate synthetic population. The main idea is to combine census micro-data with available up-to-date aggregate data. Both data sources are used to generate a set of agents for which i) the distribution and correlation of the agents' attributes are similar to those in the census micro-sample and ii) the number of agents within each category matches the current aggregate data.
By our knowledge above mentioned approach represents innovative and until now not used method for providing adequate data structures in Slovenia for implementation of agent-based micro-simulation models for needs of transport planning and land-use. Its implementation will have strong influence on new research fields - mainly on IPF - Iterative Proportional Fitting Methods, SRC - Synthetic Reconstruction Methods and COT - Combinatorial Optimization Techniques.
Synthetic population, generated in the framework of this research will represent first Slovene statistically correct, reliable, high quality and widely useful set of data on individuals and households (including correlations between them), which is needed for implementation of all kinds of modern micro-simulation models. Relevance of results is reflected by growing need for use of micro-simulation models, instead of four-step ones. Obtained results would enable wider use of micro-simulation models and consequently more accurate forecasts of future transport flows and at the same time also changes of priorities of transport policy measures.
For successful implementation of research will be needed cooperation with SURS (acquisition of adequate census sub-samples) and successfully completed survey for obtaining missing data sets. The rest represents processing and combining of different data sets with above mentioned methods (IPF, SRM, COT).
Research will be lead by FG UM, which has experience on the field of transport modelling (especially four-step models). Besides that it is also actively involved into development and implementation of activity-based model. It will cooperate with Geografski inštitut Antona Melika (numerous references on research of spatial mobility issues) and FGG UL.
Significance for science
Described approach represents innovative and until now not used method for providing adequate data structures for implementation of agent-based microsimulation transport models. Generating synthetic population based on spatially incomplete seed dataset will have strong influence on new further research on field of synthetic reconstruction - mainly on methods using IPF (Iterative Proportional Fitting) and specially IPU (Iterative Proportional Updating). Existence of synthetic population will enable study and further development of agentbased microsimulation models. Builded agentbased microsimulation model for Slovenia will enable testing of new ideas and findings on the fields of transportation planning and spatial planning. It will give new opportunities to research transferability of agent-based microsimulation transport models between different geographical/cultural environments.
Significance for the country
Synthetic population, generated in the framework of this research represents first Slovene statistically correct, reliable, high quality and widely useful set of data on individuals and households (including correlations between them), which is needed for implementation of modern microsimulation models. The relevance of results is reflected by growing need for use of agent-based microsimulation transport models, instead of conventional 4-step ones. Obtained results represent first adequate Slovene input data set. As such they will boost wider use of microsimulation models and consequently more accurate forecasts of future transport flows and at the same time they will significantly affect the changes of transport policy priorities and measures. Agentbased microsimulation traffic (and spatial) models will represent powerful tool for decision makers on national level (specially for Ministry of Infrastructure and Spatial Planning), regional and local level. Predictions of this kind of models are much more realistic, accurate and reliable than predictions of classical (four step models). They are useful in much wider scope than just traffic management (cleaner environment, less emissions, lower external costs, higher traffic safety, lower noise, better health, more efficient spatial planning, sustainable energy use, ...). With such tool will be easier to assess many effects and measures of policies mentioned above. It will be the special interest of those who are planning/developing future transport infrastructure and/or land use.
Most important scientific results
Annual report
2013,
2014,
2015,
final report
Most important socioeconomically and culturally relevant results
Annual report
2013,
2014,
2015,
final report
