Population synthesis with deep generative model: a joint household-individual approach

Abstract

This paper introduces two novel deep generative frameworks for synthetic population generation that jointly model household and individual attributes. In leveraging Variational Autoencoders (VAEs), we propose herein the SVAE-Pop2 method, which employs a single VAE with fixed-size padded inputs, along with the MVAE-Pop2 method, which uses dedicated models for various household sizes. Evaluated on a French household travel survey dataset, our experiments reveal that while both approaches effectively reproduce the actual population’s characteristics, MVAE-Pop2 achieves greater fidelity in joint attribute distributions. The proposed methodologies suggest improvements in agent-based simulations and urban modeling by means of generating realistic, multi-layered synthetic populations.

Keywords

Synthetic population / Machine learning / Deep generative model / Variational autoencoders / Household-individual / Two layers

References

[1] Aemmer, Z., & MacKenzie, D. (2022). Generative population synthesis for joint household and individual characteristics. Computers, Environment and Urban Systems,96, 101852. https://doi.org/10.1016/j.compenvurbsys.2022.101852

[2] Borysov, S. S., Rich, J., & Pereira, F. C. (2019). How to generate micro-agents? A deep generative modeling approach to population synthesis. Transportation Research Part C: Emerging Technologies,106, 73–97. https://doi.org/10.1016/j.trc.2019.07.006

[3] Borysov, S. S., Rich, J., & Pereira, F. C. (2019b). Scalable Population Synthesis with Deep Generative Modeling. Transportation Research Part C: Emerging Technologies, 106, 73–97. arXiv:1808.06910

[4] Bouzouina, L., Baraklianos, I., Bonnel, P., & Aissaoui, H. (2021). Renters vs owners: The impact of accessibility on residential location choice. Evidence from lyon urban area, france (1999–2013). Transport Policy, 109, 72–84. https://doi.org/10.1016/j.tranpol.2021.05.022

[5] BrückmannG, WillibaldF, BlancoV. Battery electric vehicle adoption in regions without strong policies. Transportation Research Part D: Transport and Environment, 2021, 90102615.

[6] CasatiD, MüllerK, FouriePJ, ErathA, AxhausenKW. Synthetic Population Generation by Combining a Hierarchical, Simulation-Based Approach with Reweighting by Generalized Raking. Transportation Research Record: Journal of the Transportation Research Board, 2015, 2493(1): 107-116.

[7] De RosaGH, PapaJP. A survey on text generation using generative adversarial networks. Pattern Recognition, 2021, 119108098.

[8] Département-Loire-Atlantique (2015). Enquête déplacements (2015) en Loire-Atlantique. https://data.loire-atlantique.fr/explore/dataset/224400028_enquete-deplacements-en-loire-atlantique/information/ Accessed 06 June 2023

[9] Duda, R. O., Stork, D. G., & Hart, P. E. (2001). Pattern classification (2nd ed.). Wiley.

[10] Fabrice YaméogoB, GastineauP, HankachP, VandanjonP-O. Comparing methods for generating a two-layered synthetic population. Transportation research record, 2021, 2675(1): 136-147.

[11] FarooqB, BierlaireM, HurtubiaR, FlötterödG. Simulation based population synthesis. Transportation Research Part B: Methodological, 2013, 58: 243-263.

[12] GarridoS, BorysovSS, PereiraFC, RichJ. Prediction of rare feature combinations in population synthesis: Application of deep generative modelling. Transportation Research Part C: Emerging Technologies, 2020, 120102787.

[13] Goeyvaerts, N., Santermans, E., Potter, G. E., Torneri, A., Kerckhove, K. V., Willem, L., Aerts, M., Beutels, P., & Hens, N. (2017). Household members do not contact each other at random: implications for infectious disease modelling. Proceedings of the Royal Society B: Biological Sciences,285. https://doi.org/10.1101/220202

[14] Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014). Generative Adversarial Networks. arXiv:1406.2661. Publisher: arXiv Version Number: 1.

[15] Higgins, I., Matthey, L., Pal, A., Burgess, C., Glorot, X., Botvinick, M., Mohamed, S., & Lerchner, A. (2017). beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework. In International Conference on Learning Representations. https://openreview.net/forum?id=Sy2fzU9gl Accessed 09 May 2023

[16] HintonGE, OsinderoS, TehY-W. A fast learning algorithm for deep belief nets. Neural computation, 2006, 18(7): 1527-1554.

[17] JohnsenM, BrandtO, GarridoS, PereiraF. Population synthesis for urban resident modeling using deep generative models. Neural Computing and Applications, 2022, 34(6): 4677-4692.

[18] KongJ, KimJ, BaeJ. Hifi-gan: Generative adversarial networks for efficient and high fidelity speech synthesis. Advances in neural information processing systems, 2020, 33: 17022-17033

[19] PlautPO. The intra-household choices regarding commuting and housing. Transportation Research Part A: Policy and Practice, 2006, 40(7): 561-571.

[20] Qian, X., Gangwal, U., Dong, S., & Davidson, R. (2024). A Deep Generative Framework for Joint Households and Individuals Population Synthesis. arXiv:2407.01643

[21] SanéAR, VandanjonP-O, BelaroussiR, HankachP. A comprehensive investigation of variational auto-encoders for population synthesis. Journal of Computational Social Science, 2025, 8113.

[22] SunL, ErathA, CaiM. A hierarchical mixture modeling framework for population synthesis. Transportation Research Part B: Methodological, 2018, 114: 199-212.

[23] SunZ, ZachariasJ. Do housing tenure and public transport provision matter in automobile use in bedroom suburban communities? evidence from beijing. Journal of Housing and the Built Environment, 2021, 36(1): 241-262.

[24] Van den Oord, A., Kalchbrenner, N., Espeholt, L., Vinyals, O., Graves, A., & et al. (2016). Conditional image generation with pixelcnn decoders. Advances in neural information processing systems, 29. https://proceedings.neurips.cc/paper_files/paper/2016/file/b1301141feffabac455e1f90a7de2054-Paper.pdf. Accessed 04 June 2024

[25] WangD, CaoX. Impacts of the built environment on activity-travel behavior: Are there differences between public and private housing residents in hong kong?. Transportation Research Part A: Policy and Practice, 2017, 103: 25-35.

[26] WangX, YanX, ZhaoX, CaoZ. Identifying latent shared mobility preference segments in low-income communities: Ride-hailing, fixed-route bus, and mobility-on-demand transit. Travel Behaviour and Society, 2022, 26: 134-142.

[27] Yameogo, B. F., Vandanjon, P.-O., Gastineau, P., & Hankach, P. (2021). Generating a Two-Layered Synthetic Population for French Municipalities: Results and Evaluation of Four Synthetic Reconstruction Methods. Journal of Artificial Societies and Social Simulation, 24(2), 5. https://doi.org/10.18564/jasss.4482

[28] Zeng, X., Wang, F., Luo, Y., gu Kang, S., Tang, J., Lightstone, F. C., Fang, E. F., Cornell, W., Nussinov, R., & Cheng, F. (2022). Deep generative molecular design reshapes drug discovery. Cell Reports Medicine, 3(12), 100794. https://doi.org/10.1016/j.xcrm.2022.100794

RIGHTS & PERMISSIONS

The Author(s)