Feasible and Novel Synthetic Population Generation with Tabular and Sequential Travel Attributes

Farbod Abbasi; Zachary Patterson; Bilal Farooq

cs.LGAug 16, 2026

Feasible and Novel Synthetic Population Generation with Tabular and Sequential Travel Attributes

Authors: Farbod Abbasi, Zachary Patterson, Bilal Farooq

Organizations: Concordia University, Montreal, Quebec, Canada · Toronto Metropolitan University, Toronto, Ontario, Canada

Abstract

Synthetic populations are critical inputs for activity-based travel demand models, yet generating realistic populations from limited survey data remains challenging. Small samples miss valid attribute combinations, known as sampling zeros, and generative models may also produce infeasible structural zeros. Moreover, realistic synthetic populations must capture both static socio-demographic attributes and sequential travel behaviour, such as trip chains. This paper proposes a regularized two-stage generative framework to address these challenges, where regularization refers to additional loss terms that guide the generator toward broader valid coverage and fewer infeasible samples. In Stage 1, a Wasserstein GAN with gradient penalty is augmented with three regularization terms, IGP, LDR, and CLAP, to improve feasibility, diversity, and novelty in tabular population synthesis. In Stage 2, Transformer and LSTM-Attention models generate sequential travel attributes, including departure time, trip purpose, and travel mode, conditioned on the synthesized tabular profiles. We also introduce novelty and count-aware metrics to evaluate whether valid unseen combinations are recovered and generated in realistic proportions. Results show that regularized models outperform the vanilla WGAN-GP across feasibility, diversity, and novelty. Regularization increases feasibility by 2.1 to 3.7 percentage points and novelty by 6.6 to 10.0 percentage points, improving sampling-zero recovery without sacrificing feasibility. The F1 score improves by 6.3 to 8.6 percentage points. For sequential attributes, LSTM-Attention best matches the trip-length distribution, while Transformer achieves higher overall sequential F1, 90.6% versus 89.1%. Cross-stage validation confirms strong consistency between generated mobility status and generated trip chains.

Figures & tables

Study	Model	Seq.	Samp. Zeros	Struct. Zeros	Grp.	Eval.
Traditional Methods
Zhu and Ferreira Jr (2014) ; Prédhumeau and Manley (2023)	IPF / QISI	$\times$	$\times$	$\times$	Trad.	$\times$
Farooq et al. (2013) ; Saadi et al. (2016) ; Kukic et al. (2024)	MCMC / HMM	$\times$	$\times$	$\times$	Trad.	$\times$
Sun and Erath (2015) ; Rahman and Fatmi (2023)	Bayesian Network	$\times$	$\times$	$\times$	Trad.	$\times$
Sun et al. (2018) ; Kukic et al. (2024)	Hierarchical / Multilevel	$\times$	$\times$	$\times$	Trad.	$\times$
Hörl and Balac (2021) ; Fournier et al. (2021)	Integrated Pipeline	$\times$	$\times$	$\times$	Trad.	$\times$

Table 1: Summary of related work on synthetic population generation.

\checkmark

= addressed,

\times

= not addressed,

\sim

= partially addressed. Samp. = Sampling Zeros; Struct. = Structural Zeros; Seq. = Sequential generation; Eval. = Evaluation completeness (novelty and count-aware metrics).

Attribute (Dimensions)	Category	Proportion (%)
1. Number of vehicles (m_auto)	0	9.49
	1	36.41
	2	39.95
	3+	14.15
2. Household size (m_pers)	1	13.03
	2	34.92

Table 2: Tabular socio-demographic attributes and category proportions in the 2018 Montreal OD dataset.

Figure 1: Trip Sequence Length Distribution

Figure 2: Relationship between sample size and sampling zeros

Figure 3: Two stage framework for synthetic population generation

Figure 4: Overview of the proposed regularized WGAN-GP framework for tabular attribute synthesis

Panel of Figure 5 — Figure 5: Overview of the evaluation framework for assessing tabular quality, sequential quality, and cross-stage consistency.

Figure 6: Comparison of marginal probability distributions.

Figure 7: Mean SRMSE across univariate, bivariate, and trivariate distributions for all model variants.

Method	$n_{\text{real}}$	$n_{\text{sample}}$	$n_{\text{unseen}}$	$n_{\text{fake}}$	Feasibility	Diversity	Div. (w)	Novelty	Nov. (w)
Vanilla	4282	628	3654	6351	0.898	0.638	0.666	0.600	0.479
CLAP				5412	0.935	0.678	0.684	0.637	0.548
IGP				6408	0.919	0.741	0.684	0.706	0.579
LDR				5460	0.934	0.714	0.690	0.676	0.545

Table 3: Combination-level validity, diversity, and novelty results for the generated datasets.

Method	$F1_{D}$	$F1_{D_{w}}$	$F1_{N}$	$F1_{N_{w}}$
Vanilla	0.746	0.765	0.719	0.625
CLAP	0.786	0.790	0.758	0.691
IGP	0.820	0.784	0.799	0.711
LDR	0.809	0.793	0.784	0.688

Table 4: F1-based evaluation combining feasibility with diversity and novelty metrics.

Figure 8: Trip-length distribution comparison between the real OD data and the generated outputs.

Figure 9: Distribution of the ten most frequent real activity sequences and their corresponding shares in the CLAP-LSTM generated population.

Figure 10: N-gram evaluation of LSTM-Attention and Transformer. The shaded area shows the range of results across the Stage 1 tabular model variants.

Feature	Method	N-gram	LSTM			Transformer
Feature	Method	N-gram	Diversity	Novelty	Feasibility	Diversity	Novelty	Feasibility
d_grhre_vec	vanilla	unigram	1.000	–	1.000	1.000	–	1.000
		bigram	1.000	1.000	0.969	1.000	1.000	0.968
		trigram	0.911	0.781	0.907	0.949	0.875	0.906
	IGP	unigram	1.000	–	1.000	1.000	–	1.000
		bigram	1.000	1.000	0.969	1.000	1.000	0.965

Table 5: Combined N-gram evaluation results for all features using a 1% sample. Results are shown separately for each feature and model family.

Figure 11: Overall sequential quality ranking of the eight model combinations based on the mean F1 score.

Method	LSTM				Transformer
Method	Violations all 3 = [0]	Violations all 3 $\neq$ [0]	Total violations	Accuracy (%)	Violations all 3 = [0]	Violations all 3 $\neq$ [0]	Total violations	Accuracy (%)
IGP	2278	16	2294	98.59	771	2	773	99.52
LDR	2804	9	2813	98.27	647	5	652	99.60
vanilla	1343	22	1365	99.13	466	8	474	99.71
CLAP	1089	17	1106	99.32	316	5	321	99.80

Table 6: Consistency check between p_mobil and the three trip-related features.

Explore similar work

Feb 17, 2026

Enhancing Diversity and Feasibility: Joint Population Synthesis from Multi-source Data Using Generative Models

Generating realistic synthetic populations is essential for agent-based models (ABM) in transportation and urban planning. Current methods face two major limitations. First, many rely on a single dataset or follow a sequential data fusion and generation process, which means they fail to capture the complex interplay between features. Second, these approaches struggle with sampling zeros (valid but unobserved attribute combinations) and structural zeros (infeasible combinations due to logical constraints), which reduce the diversity and feasibility of the generated data. This study proposes a novel method to simultaneously integrate and synthesize multi-source datasets using a Wasserstein Generative Adversarial Network (WGAN) with gradient penalty. This joint learning method improves both the diversity and feasibility of synthetic data by defining a regularization term (inverse gradient penalty) for the generator loss function. For the evaluation, we implement a unified evaluation metric for similarity, and place special emphasis on measuring diversity and feasibility through recall, precision, and the F1 score. Results show that the proposed joint approach outperforms the sequential baseline, with recall increasing by 7% and precision by 15%. Additionally, the regularization term further improves diversity and feasibility, reflected in a 10% increase in recall and 1% in precision. We assess similarity distributions using a five-metric score. The joint approach performs better overall, and reaches a score of 88.1 compared to 84.6 for the sequential method. Since synthetic populations serve as a key input for ABM, this multi-source generative approach has the potential to significantly enhance the accuracy and reliability of ABM.

Farbod Abbasi, Zachary Patterson, Bilal Farooq

Concordia University, Montreal, QC, Canada · Concordia Institute for Information Systems Engineering, Concordia University, Montreal, QC, Canada · Toronto Metropolitan University, Toronto, ON, Canada

Synthetic Data Generation Synthetic Tabular Data Generation Constrained Generative Modeling Generative Adversarial Networks

Ask PDF

Sep 8, 2026

LEBGen: An LLM-Enhanced Bayesian Network Framework for Few-Shot Travel Survey Data Generation

Travel survey data are essential for transportation planning and travel behavior analysis, yet collecting large-scale representative samples is costly and time-consuming. A practical alternative is to generate synthetic survey records from a few-shot sample. However, such samples provide incomplete coverage of heterogeneous traveler groups and insufficient evidence for recovering the complex dependencies between demographic characteristics and travel behavior. Existing approaches have complementary limitations. Probabilistic generative models such as Bayesian networks (BNs) offer explicit distributional control, but structures learned from few-shot samples may omit meaningful dependencies or retain spurious ones. Large language models (LLMs) can help address these difficulties in BN structure learning by providing behavioral knowledge that complements the limited statistical evidence. We therefore propose LEBGen, an LLM-enhanced BN framework that uses this knowledge to refine network structure for few-shot travel survey data generation. Specifically, the LLM first identifies traveler personas from demographic attribute and travel behavior statistics, then recovers dependencies missed by the persona-augmented BN structure and prune spurious ones. The refined BN is parameterized exclusively from the observed data to generate synthetic records. Under a 2% few-shot setting on the 2022 Hong Kong Travel Characteristics Survey, LEBGen reduces the mean marginal Jensen-Shannon divergence from 0.0671 to 0.0091 and the mean absolute Cramer's V error by 14.3% over the best-performing baseline, substantially improving both distributional and dependency fidelity.

Zijian Shen, Bin Zhou, Jiguang Wang +2

Department of Civil Engineering, The University of Hong Kong · Department of Urban Planning and Design, The University of Hong Kong

Bayesian Networks Synthetic Data Generation Synthetic Tabular Data Generation

Ask PDF

Jun 10, 2026

TrajGenAgent: A Hierarchical LLM Agent for Human Mobility Trajectory Generation

Human mobility data is important for transportation, urban planning, and epidemic control, but large-scale trajectory collection is often costly and privacy-constrained, motivating realistic synthetic trajectory generation. Existing LLM-based generators typically rely on either prompt engineering, which preserves zero-shot reasoning but lacks fine-grained spatiotemporal grounding, or trajectory-level fine-tuning, which improves statistical precision but incurs substantial computational cost and may weaken general reasoning. We propose TrajGenAgent, a semantic-aware hierarchical LLM-agent framework for human mobility trajectory generation without model fine-tuning. TrajGenAgent uses a two-stage orchestrator-worker design: an LLM first synthesizes an individual- and weekday-conditioned activity chain from historical evidence via in-context learning, and a deterministic workflow then grounds each activity into a complete visit using personalized POI retrieval, distance-aware location selection, kinematics-aware travel-time propagation, and LLM-based duration estimation. To evaluate realism beyond aggregate spatiotemporal statistics, we introduce an anomaly-detection-based evaluation framework using two complementary detectors to assess behavioral and semantic plausibility. Experiments on benchmark and large-scale simulation datasets show that TrajGenAgent improves spatiotemporal fidelity, semantic coherence, and individual-specific behavioral realism over representative neural and LLM-based baselines, while avoiding parameter updates.

Siyu Li, Toan Tran, Lingyi Zhao +2

Dept. of Computer Science, Emory University, Atlanta, GA, USA · Novateur Research Solutions, Ashburn, VA, USA

Trajectory Generation Human Behavior Simulation Synthetic Data Generation LLM Agent Trajectory Synthesis LLM Agent Planning

Ask PDF