Academic Research

In this section

Statistical Modelling and Data Science Group (SMDSG)

Advancing statistical modelling, causal inference, and data-centric engineering for resilient and intelligent transport systems.

Research Mission

The TSC Statistical Modelling and Data Science Group (SMDSG) conducts both methodological and applied work at the interface of transport systems, causal inference, and data-centric engineering.

Our mission is to advance quantitative methods that support:

safe, efficient, and resilient transport operations;
adaptive intelligent infrastructure systems; and
evidence-based transport interventions.

About the Group

The SMDSG is led by , Head of CTEM and Co-Director of TSC.

Our research combines methodological innovation in statistics, causal inference, data-centric engineering and machine learning with applied work on transport operations, safety, and resilience.

We develop quantitative tools that support the safe, efficient, and sustainable management of transport systems.

Our work is organised around four main research themes:

Statistical Modelling and Network Performance
Data-Centric Engineering and Complex Systems
Resilience, Safety, and Risk
Causal inference and prediction

Recent Publications

Explore our group’s latest research outputs and publications here.

Research Areas

The SMDSG's work contributes to the next generation of intelligent, data-informed transport systems, integrating statistical science with engineering applications to make transport safer, more resilient, and more efficient.

Statistical Modelling and Network Performance Analytics

We develop and apply advanced statistical and data science models to analyse and predict the performance of mass transport networks using large-scale, high-dimensional data.

Our work encompasses causal inference, stochastic network modelling, Bayesian estimation, and efficiency frontier analysis, providing tools to benchmark and optimise system performance.

Applications include metro congestion modelling, journey-time variance decomposition, cost comparisons, and elasticity estimation, supporting operational strategy, investment appraisal, and transport policy evaluation.

Data-Centric Engineering and Complex Systems

Our research at the intersection of engineering, data science, and statistical learning, develops data-centric methods for monitoring, control, and optimisation of complex transport infrastructures.

We integrate sensor, smart-card, and telemetry data with machine learning interpretability, probabilistic programming, and scalable causal estimation to detect anomalies, forecast disruptions, and build digital twins of network performance.

Applications include fault detection, performance monitoring, and predictive maintenance, underpinning collaborations with global transit operators to design

Resilience, Safety, and Risk in Transport Systems

We study transport system resilience, safety, and risk using causal and predictive models, quantifying vulnerability, failure propagation, and exposure to shocks from operational perturbations to pandemics.

Recent work applies Bayesian and neural models to analyse safety interventions, weather-related pedestrian risk, and COVID-19 transit impacts. We identify latent risk drivers, supporting proactive safety management and the design of robust, adaptive control strategies.

We also develop frameworks linking infrastructure performance, system attributes, and environmental features.

Causal Inference and Prediction

We advance causal inference methodologies for transport and spatial systems, bridging predictive analytics with intervention evaluation.

Our research has developed fundamental methods for doubly robust estimation, quasi-experimental designs, and Bayesian structural modelling to identify cause–effect relationships in complex, observational data.

These frameworks inform studies of infrastructure investment impacts, pricing strategies, and mobility policies, complemented by explainable AI and ensemble machine learning to support decision-making under uncertainty.