arXiv subjects

Prateek Bansal

Publications and source records attributed to Prateek Bansal.

At least 19 recordsLinked to original sources

arXiv2026

Deep adaptive design with an evidential bias criterion

Bayesian optimal experimental design (BOED) aims to collect informative data by optimizing an expected utility reflecting the goals of an experiment. However, this optimization is computationally challenging for common utilities and complex models. This is especially so for sequential or adaptive designs, where design and data collection alternate, so that feedback from already observed data must be taken into account. Most existing BOED research employs information gain as the utility, leading to the expected information gain (EIG) criterion. While EIG is widely useful, it may not always adequately reflect experimental goals. EIG can be viewed as rewarding experiments that produce large positive evidence for the truth on average, but it does not directly control the risk of an experiment producing misleading evidence. Here we consider an alternative criterion, which we call bias against (BA), that prioritizes such control. To address computational challenges when applying this criterion for adaptive design, we consider a policy-based deep adaptive design framework, which has previously been used for the EIG criterion. Minimizing a tractable upper bound on the BA objective is equivalent to maximizing a variance-penalized EIG criterion, and we optimize the latter by approximating it by Monte Carlo and learning design policies using stochastic gradient methods. The differences between BA and EIG designs are demonstrated in several examples including the adaptive design of a complex discrete choice experiment.

stat.ME↗

arXiv2025

Physics Informed Multi-task Joint Generative Learning for Arterial Vehicle Trajectory Reconstruction Considering Lane Changing Behavior

Reconstructing complete traffic flow time-space diagrams from vehicle trajectories offer a comprehensive view on traffic dynamics at arterial intersections. However, obtaining full trajectories across networks is costly, and accurately inferring lane-changing (LC) and car-following behaviors in multi-lane environments remains challenging. This study proposes a generative framework for arterial vehicle trajectory reconstruction that jointly models lane-changing and car-following behaviors through physics-informed multi-task joint learning. The framework consists of a Lane-Change Generative Adversarial Network (LC-GAN) and a Trajectory-GAN. The LC-GAN models stochastic LC behavior from historical trajectories while considering physical conditions of arterial intersections, such as signal control, geometric configuration, and interactions with surrounding vehicles. The Trajectory-GAN then incorporates LC information from the LC-GAN with initial trajectories generated from physics-based car-following models, refining them in a data-driven manner to adapt to dynamic traffic conditions. The proposed framework is designed to reconstruct complete trajectories from only a small subset of connected vehicle (CV) trajectories; for example, even a single observed trajectory per lane, by incorporating partial trajectory information into the generative process. A multi-task joint learning facilitates synergistic interaction between the LC-GAN and Trajectory-GAN, allowing each component to serves as both auxiliary supervision and a physical condition for the other. Validation using two real-world trajectory datasets demonstrates that the framework outperforms conventional benchmark models in reconstructing complete time-space diagrams for multi-lane arterial intersections. This research advances the integration of trajectory-based sensing from CVs with physics-informed deep learning.

eess.SY↗

arXiv2025

Large Language Models Enable Design of Personalized Nudges across Cultures

Nudge strategies are effective tools for influencing behaviour, but their impact depends on individual preferences. Strategies that work for some individuals may be counterproductive for others. We hypothesize that large language models (LLMs) can facilitate the design of individual-specific nudges without the need for costly and time-intensive behavioural data collection and modelling. To test this, we use LLMs to design personalized decoy-based nudges tailored to individual profiles and cultural contexts, aimed at encouraging air travellers to voluntarily offset CO$_2$ emissions from flights. We evaluate their effectiveness through a large-scale survey experiment ($n=3495$) conducted across five countries. Results show that LLM-informed personalized nudges are more effective than uniform settings, raising offsetting rates by 3-7$\%$ in Germany, Singapore, and the US, though not in China or India. Our study highlights the potential of LLM as a low-cost testbed for piloting nudge strategies. At the same time, cultural heterogeneity constrains their generalizability underscoring the need for combining LLM-based simulations with targeted empirical validation.

cs.CY↗

arXiv2025

Causal resilience curves: A data-driven framework for quantifying the spatiotemporal impacts of metro service disruptions

Urban metro systems move vast numbers of passengers with a high level of efficiency in resource use, but frequently experience disruptions that result in delays, crowding, and deterioration in passenger satisfaction and patronage. To quantify these adverse consequences, this paper presents a novel, data-driven causal inference framework to measure metro resilience by estimating both the direct and spillover effects of service disruptions on passenger demand, journey time, travel speed and on-board crowding. By integrating high-frequency smart card data into a synthetic control design, we use weighted non-disrupted days to construct unbiased counterfactuals, which resolves confounding factors and accurately captures disruption propagation across the network. The impact estimates are further translated into station-level causal resilience curves that reveal spatial heterogeneity in the temporal patterns of degradation and recovery across locations, providing metro operators with actionable insights for targeted interventions and resource allocation. A case study of the Hong Kong MTR demonstrates the framework's superiority over naive typical-day comparisons and machine-learning benchmarks in delivering unbiased resilience curves. This paper is the first to derive causal estimates of dynamic metro resilience. This practical tool can be generalised to evaluate resilience in a broad range of public transport systems.

stat.AP↗

arXiv2025

Scalable Variational Inference for Multinomial Probit Models under Large Choice Sets and Sample Sizes

The multinomial probit (MNP) model is widely used to analyze categorical outcomes due to its ability to capture flexible substitution patterns among alternatives. Conventional likelihood based and Markov chain Monte Carlo (MCMC) estimators become computationally prohibitive in high dimensional choice settings. This study introduces a fast and accurate conditional variational inference (CVI) approach to calibrate MNP model parameters, which is scalable to large samples and large choice sets. A flexible variational distribution on correlated latent utilities is defined using neural embeddings, and a reparameterization trick is used to ensure the positive definiteness of the resulting covariance matrix. The resulting CVI estimator is similar to a variational autoencoder, with the variational model being the encoder and the MNP's data generating process being the decoder. Straight through estimation and Gumbel SoftMax approximation are adopted for the argmax operation to select an alternative with the highest latent utility. This eliminates the need to sample from high dimensional truncated Gaussian distributions, significantly reducing computational costs as the number of alternatives grows. The proposed method achieves parameter recovery comparable to MCMC. It can calibrate MNP parameters with 20 alternatives and one million observations in approximately 28 minutes roughly 36 times faster and more accurate than the existing benchmarks in recovering model parameters.

stat.ME↗

arXiv2025

A Large Language Model for Feasible and Diverse Population Synthesis

Generating a synthetic population that is both feasible and diverse is crucial for ensuring the validity of downstream activity schedule simulation in activity-based models (ABMs). While deep generative models (DGMs), such as variational autoencoders and generative adversarial networks, have been applied to this task, they often struggle to balance the inclusion of rare but plausible combinations (i.e., sampling zeros) with the exclusion of implausible ones (i.e., structural zeros). To improve feasibility while maintaining diversity, we propose a fine-tuning method for large language models (LLMs) that explicitly controls the autoregressive generation process through topological orderings derived from a Bayesian Network (BN). Experimental results show that our hybrid LLM-BN approach outperforms both traditional DGMs and proprietary LLMs (e.g., ChatGPT-4o) with few-shot learning. Specifically, our approach achieves approximately 95% feasibility, significantly higher than the ~80% observed in DGMs, while maintaining comparable diversity, making it well-suited for practical applications. Importantly, the method is based on a lightweight open-source LLM, enabling fine-tuning and inference on standard personal computing environments. This makes the approach cost-effective and scalable for large-scale applications, such as synthesizing populations in megacities, without relying on expensive infrastructure. By initiating the ABM pipeline with high-quality synthetic populations, our method improves overall simulation reliability and reduces downstream error propagation. The source code for these methods is available for research and practical application.

cs.LG↗

arXiv2025

Variational inference for hierarchical models with conditional scale and skewness corrections

Gaussian variational approximations are widely used for summarizing posterior distributions in Bayesian models, especially in high-dimensional settings. However, a drawback of such approximations is the inability to capture skewness or more complex features of the posterior. Recent work suggests applying skewness corrections to existing Gaussian or other symmetric approximations to address this limitation. We propose to incorporate the skewness correction into the definition of an approximating variational family. We consider approximating the posterior for hierarchical models, in which there are ``global'' and ``local'' parameters. A baseline variational approximation is defined as the product of a Gaussian marginal posterior for global parameters and a Gaussian conditional posterior for local parameters given the global ones. Skewness corrections are then considered. The adjustment of the conditional posterior term for local variables is adaptive to the global parameter value. Optimization of baseline variational parameters is performed jointly with the skewness correction. Our approach allows the location, scale and skewness to be captured separately, without using additional parameters for skewness adjustments. The proposed method substantially improves accuracy for only a modest increase in computational cost compared to state-of-the-art Gaussian approximations. Good performance is demonstrated in generalized linear mixed models and multinomial logit discrete choice models.

stat.ME↗

arXiv2024

Multi-objective Bayesian optimization for Likelihood-Free inference in sequential sampling models of decision making

Statistical models are often defined by a generative process for simulating synthetic data, but this can lead to intractable likelihoods. Likelihood free inference (LFI) methods enable Bayesian inference to be performed in this case. Extending a popular approach to simulation-efficient LFI for single-source data, we propose Multi-objective Bayesian Optimization for Likelihood Free Inference (MOBOLFI) to perform LFI using multi-source data. MOBOLFI models a multi-dimensional discrepancy between observed and simulated data, using a separate discrepancy for each data source. The use of a multivariate discrepancy allows for approximations to individual data source likelihoods in addition to the joint likelihood, enabling detection of conflicting information and deeper understanding of the importance of different data sources in estimating individual parameters. The adaptive choice of simulation parameters using multi-objective Bayesian optimization ensures simulation efficient approximation of likelihood components for all data sources. We illustrate our approach in sequential sampling models (SSMs), which are widely used in psychology and consumer-behavior modeling. SSMs are often fitted using multi-source data, such as choice and response time. The advantages of our approach are illustrated in comparison with a single discrepancy for an SSM fitted to data assessing preferences of ride-hailing drivers in Singapore to rent electric vehicles.

stat.ME↗

arXiv2023

Revisiting McFadden's correction factor for sampling of alternatives in multinomial logit and mixed multinomial logit models

In this paper, we revisit McFadden (1978)'s correction factor for sampling of alternatives in multinomial logit (MNL) and mixed multinomial logit (MMNL) models. McFadden (1978) proved that consistent parameter estimates are obtained when estimating MNL models using a sampled subset of alternatives, including the chosen alternative, in combination with a correction factor. We decompose this correction factor into i) a correction for overestimating the MNL choice probability due to using a smaller subset of alternatives, and ii) a correction for which a subset of alternatives is contrasted through utility differences and thereby the extent to which we learn about the parameters of interest in MNL. Keane and Wasi (2016) proved that the overall expected positive information divergence - comprising the above two elements - is minimised between the true and sampled likelihood when applying a sampling protocol satisfying uniform conditioning. We generalise their result to the case of positive conditioning and show that whilst McFadden (1978)'s correction factor may not minimise the overall expected information divergence, it does minimise the expected information loss with respect to the parameters of interest. We apply this result in the context of Bayesian analysis and show that McFadden (1978)'s correction factor minimises the expected information loss regarding the parameters of interest across the entire posterior density irrespective of sample size. In other words, McFadden (1978)'s correction factor has desirable small and large sample properties. We also show that our results for Bayesian MNL models transfer to MMNL and that only McFadden (1978) correction factor is sufficient to minimise the expected information loss in the parameters of interest. Monte Carlo simulations illustrate the successful application of sampling of alternatives in Bayesian MMNL models.

stat.ME↗

arXiv2023

Nonparametric mixed logit model with market-level parameters estimated from market share data

We propose a nonparametric mixed logit model that is estimated using market-level choice share data. The model treats each market as an agent and represents taste heterogeneity through market-specific parameters by solving a multiagent inverse utility maximization problem, addressing the limitations of existing market-level choice models with parametric estimation. A simulation study is conducted to evaluate the performance of our model in terms of estimation time, estimation accuracy, and out-of-sample predictive accuracy. In a real data application, we estimate the travel mode choice of 53.55 million trips made by 19.53 million residents in New York State. These trips are aggregated based on population segments and census block group-level origin-destination (OD) pairs, resulting in 120,740 markets. We benchmark our model against multinomial logit (MNL), nested logit (NL), inverse product differentiation logit (IPDL), and the BLP models. The results show that the proposed model improves the out-of-sample accuracy from 65.30% to 81.78%, with a computation time less than one-tenth of that taken to estimate the BLP model. The price elasticities and diversion ratios retrieved from our model and benchmark models exhibit similar substitution patterns. Moreover, the market-level parameters estimated by our model provide additional insights and facilitate their seamless integration into supply-side optimization models for transportation design. By measuring the compensating variation for the driving mode, we found that a $9 congestion toll would impact roughly 60 % of the total travelers. As an application of supply-demand integration, we showed that a 50% discount of transit fare could bring a maximum ridership increase of 9402 trips per day under a budget of $50,000 per day.

econ.EM↗

arXiv2022

A Data Fusion Approach for Ride-sourcing Demand Estimation: A Discrete Choice Model with Sampling and Endogeneity Corrections

Ride-sourcing services offered by companies like Uber and Didi have grown rapidly in the last decade. Understanding the demand for these services is essential for planning and managing modern transportation systems. Existing studies develop statistical models for ride-sourcing demand estimation at an aggregate level due to limited data availability. These models lack foundations in microeconomic theory, ignore competition of ride-sourcing with other travel modes, and cannot be seamlessly integrated into existing individual-level (disaggregate) activity-based models to evaluate system-level impacts of ride-sourcing services. In this paper, we present and apply an approach for estimating ride-sourcing demand at a disaggregate level using discrete choice models and multiple data sources. We first construct a sample of trip-based mode choices in Chicago, USA by enriching household travel survey with publicly available ride-sourcing and taxi trip records. We then formulate a multivariate extreme value-based discrete choice with sampling and endogeneity corrections to account for the construction of the estimation sample from multiple data sources and endogeneity biases arising from supply-side constraints and surge pricing mechanisms in ride-sourcing systems. Our analysis of the constructed dataset reveals insights into the influence of various socio-economic, land use and built environment features on ride-sourcing demand. We also derive elasticities of ride-sourcing demand relative to travel cost and time. Finally, we illustrate how the developed model can be employed to quantify the welfare implications of ride-sourcing policies and regulations such as terminating certain types of services and introducing ride-sourcing taxes.

econ.EM↗

arXiv2022

Robust discrete choice models with t-distributed kernel errors

Outliers in discrete choice response data may result from misclassification and misreporting of the response variable and from choice behaviour that is inconsistent with modelling assumptions (e.g. random utility maximisation). In the presence of outliers, standard discrete choice models produce biased estimates and suffer from compromised predictive accuracy. Robust statistical models are less sensitive to outliers than standard non-robust models. This paper analyses two robust alternatives to the multinomial probit (MNP) model. The two models are robit models whose kernel error distributions are heavy-tailed t-distributions to moderate the influence of outliers. The first model is the multinomial robit (MNR) model, in which a generic degrees of freedom parameter controls the heavy-tailedness of the kernel error distribution. The second model, the generalised multinomial robit (Gen-MNR) model, is more flexible than MNR, as it allows for distinct heavy-tailedness in each dimension of the kernel error distribution. For both models, we derive Gibbs samplers for posterior inference. In a simulation study, we illustrate the excellent finite sample properties of the proposed Bayes estimators and show that MNR and Gen-MNR produce more accurate estimates if the choice data contain outliers through the lens of the non-robust MNP model. In a case study on transport mode choice behaviour, MNR and Gen-MNR outperform MNP by substantial margins in terms of in-sample fit and out-of-sample predictive accuracy. The case study also highlights differences in elasticity estimates across models.

econ.EM↗

arXiv2022

A Deep Generative Model for Feasible and Diverse Population Synthesis

An ideal synthetic population, a key input to activity-based models, mimics the distribution of the individual- and household-level attributes in the actual population. Since the entire population's attributes are generally unavailable, household travel survey (HTS) samples are used for population synthesis. Synthesizing population by directly sampling from HTS ignores the attribute combinations that are unobserved in the HTS samples but exist in the population, called 'sampling zeros'. A deep generative model (DGM) can potentially synthesize the sampling zeros but at the expense of generating 'structural zeros' (i.e., the infeasible attribute combinations that do not exist in the population). This study proposes a novel method to minimize structural zeros while preserving sampling zeros. Two regularizations are devised to customize the training of the DGM and applied to a generative adversarial network (GAN) and a variational autoencoder (VAE). The adopted metrics for feasibility and diversity of the synthetic population indicate the capability of generating sampling and structural zeros -- lower structural zeros and lower sampling zeros indicate the higher feasibility and the lower diversity, respectively. Results show that the proposed regularizations achieve considerable performance improvement in feasibility and diversity of the synthesized population over traditional models. The proposed VAE additionally generated 23.5% of the population ignored by the sample with 79.2% precision (i.e., 20.8% structural zeros rates), while the proposed GAN generated 18.3% of the ignored population with 89.0% precision. The proposed improvement in DGM generates a more feasible and diverse synthetic population, which is critical for the accuracy of an activity-based model.

stat.ML↗

arXiv2022

DT2I: Dense Text-to-Image Generation from Region Descriptions

Despite astonishing progress, generating realistic images of complex scenes remains a challenging problem. Recently, layout-to-image synthesis approaches have attracted much interest by conditioning the generator on a list of bounding boxes and corresponding class labels. However, previous approaches are very restrictive because the set of labels is fixed a priori. Meanwhile, text-to-image synthesis methods have substantially improved and provide a flexible way for conditional image generation. In this work, we introduce dense text-to-image (DT2I) synthesis as a new task to pave the way toward more intuitive image generation. Furthermore, we propose DTC-GAN, a novel method to generate images from semantically rich region descriptions, and a multi-modal region feature matching loss to encourage semantic image-text matching. Our results demonstrate the capability of our approach to generate plausible images of complex scenes using region captions.

cs.CV↗

arXiv2022

Optimal congestion control strategies for near-capacity urban metros: informing intervention via fundamental diagrams

Congestion; operational delays due to a vicious circle of passenger-congestion and train-queuing; is an escalating problem for metro systems because it has negative consequences from passenger discomfort to eventual mode-shifts. Congestion arises due to large volumes of passenger boardings and alightings at bottleneck stations, which may lead to increased stopping times at stations and consequent queuing of trains upstream, further reducing line throughput and implying an even greater accumulation of passengers at stations. Alleviating congestion requires control strategies such as regulating the inflow of passengers entering bottleneck stations. The availability of large-scale smartcard and train movement data from day-to-day operations facilitates the development of models that can inform such strategies in a data-driven way. In this paper, we propose to model station-level passenger-congestion via empirical passenger boarding-alightings and train flow relationships, henceforth, fundamental diagrams (FDs). We emphasise that estimating FDs using station-level data is empirically challenging due to confounding biases arising from the interdependence of operations at different stations, which obscures the true sources of congestion in the network. We thus adopt a causal statistical modelling approach to produce FDs that are robust to confounding and as such suitable to properly inform control strategies. The closest antecedent to the proposed model is the FD for road traffic networks, which informs traffic management strategies, for instance, via locating the optimum operation point. Our analysis of data from the Mass Transit Railway, Hong Kong indicates the existence of concave FDs at identified bottleneck stations, and an associated critical level of boarding-alightings above which congestion sets-in unless there is an intervention.

stat.AP↗

arXiv2022

Fuel consumption elasticities, rebound effect and feebate effectiveness in the Indian and Chinese new car markets

China and India, the world's two most populous developing economies, are also among the world's largest automotive markets and carbon emitters. To reduce carbon emissions from the passenger car sector, both countries have considered various policy levers affecting fuel prices, car prices and fuel economy. This study estimates the responsiveness of new car buyers in China and India to such policy levers and drivers including income. Furthermore, we estimate the potential for rebound effect and the effectiveness of a feebate policy. To accomplish this, we developed a joint discrete-continuous model of car choice and usage based on revealed preference survey data from approximately 8000 new car buyers from India and China who purchased cars in 2016-17. Conditional on buying a new car, the fuel consumption in both markets is found to be relatively unresponsive to fuel price and income, with magnitudes of elasticity estimates ranging from 0.12 to 0.15. For both markets, the mean segment-level direct elasticities of fuel consumption relative to car price and fuel economy range from 0.57 to 0.65. The rebound effect on fuel savings due to cost-free fuel economy improvement is found to be 17.1% for India and 18.8% for China. A revenue-neutral feebate policy, with average rebates and fees of up to around 15% of the retail price, resulted in fuel savings of around 0.7% for both markets. While the feebate policy's rebound effect is low - 7.3% for India and 1.6% for China - it does not appear to be an effective fuel conservation policy.

econ.GN↗

arXiv2021

A Multinomial Probit Model with Choquet Integral and Attribute Cut-offs

Several non-linear functions and machine learning methods have been developed for flexible specification of the systematic utility in discrete choice models. However, they lack interpretability, do not ensure monotonicity conditions, and restrict substitution patterns. We address the first two challenges by modelling the systematic utility using the Choquet Integral (CI) function and the last one by embedding CI into the multinomial probit (MNP) choice probability kernel. We also extend the MNP-CI model to account for attribute cut-offs that enable a modeller to approximately mimic the semi-compensatory behaviour using the traditional choice experiment data. The MNP-CI model is estimated using a constrained maximum likelihood approach, and its statistical properties are validated through a comprehensive Monte Carlo study. The CI-based choice model is empirically advantageous as it captures interaction effects while maintaining monotonicity. It also provides information on the complementarity between pairs of attributes coupled with their importance ranking as a by-product of the estimation. These insights could potentially assist policymakers in making policies to improve the preference level for an alternative. These advantages of the MNP-CI model with attribute cut-offs are illustrated in an empirical application to understand New Yorkers' preferences towards mobility-on-demand services.

stat.AP↗

arXiv2021

A General Framework to Forecast the Adoption of Novel Products: A Case of Autonomous Vehicles

Due to the unavailability of prototypes, the early adopters of novel products actively seek information from multiple sources (e.g., media and social networks) to minimize the potential risk. The existing behavior models not only fail to capture the information propagation within the individual's social network, but also they do not incorporate the impact of such word-of-mouth (WOM) dissemination on the consumer's risk preferences. Moreover, even cutting-edge forecasting models rely on crude/synthetic consumer behavior models. We propose a general framework to forecast the adoption of novel products by developing a new consumer behavior model and integrating it into a population-level agent-based model. Specifically, we extend the hybrid choice model to estimate consumer behavior, which incorporates social network effects and interplay between WOM and risk aversion. The calibrated consumer behavior model and synthetic population are passed through the agent-based model for forecasting the product market share. We apply the proposed framework to forecast the adoption of autonomous vehicles (AVs) in Nashville, USA. The consumer behavior model is calibrated with a stated preference survey data of 1,495 Nashville residents. The output of the agent-based model provides the effect of the purchase price, post-purchase satisfaction, and safety measures/regulations on the forecasted AV market share. With an annual AV price reduction of 5% at the initial purchase price of $40,000 and 90% of satisfied adopters, AVs are forecasted to attain around 85% market share in thirty years. These findings are crucial for policymakers to develop infrastructure plans and manufacturers to conduct an after-sales cost-benefit analysis.

econ.GN↗