Fred Shone

I build tools for understanding and shaping complex human systems. My work sits at the intersection of transport modelling, generative machine learning, and activity-based simulation — translating novel research into software that practitioners actually use.

Most recently I was Technical Lead of Arup’s City Modelling Lab (CML), a specialist team advancing large-scale transport and urban modelling. I led the design and delivery of open-source modelling infrastructure used across Arup’s global practice. Read about the team’s work on the CML blog.

I am completing my PhD at UCL’s Behaviour & Infrastructure Group (BIG), where my research applies deep generative models to synthesise realistic human activity schedules — the behavioural building blocks of transport, energy, and epidemiology models. My 2025 paper in Transportation Research Part C presents a significantly faster and more flexible approach to schedule synthesis than existing methods. My next work, ActVAE, is on the way.

I am the author of pam, a Python library for population activity modelling published in the Journal of Open Source Software. I have also built mc and elara to support serious MATSim modelling workflows. I am currently working on robin and caveat, projects for population and activity schedule synthesis. I am currently starting to assemble an evaluation frameowrk starting with acteval.

I am available for senior roles, consulting, and research collaboration in transport modelling, simulation, applied ML, and building stuff. Get in touch or view my CV.

I also climb, play ultimate, and make things.

latest posts

Mar 03, 2026	The Unreasonable Difficulty of Modelling Reasonable People
Sep 26, 2025	A soft introduction to VAEs
Jul 13, 2025	Evaluation

selected publications

Synthesising activity participations and scheduling with deep generative machine learning

Fred Shone, and Tim Hillel

Transportation Research Part C: Emerging Technologies, 2025

Abs

Using a deep generative machine learning approach, we synthesise human activity participations and scheduling; i.e. the choices of what activities to participate in and when. Activity schedules are a core component of many applied transport, energy, and epidemiology models. Our data-driven approach directly learns the distributions resulting from human preferences and scheduling logic without the need for complex interacting combinations of sub-models and custom rules. This makes our approach significantly faster and simpler to operate than existing approaches to synthesise or anonymise schedule data. We additionally contribute a novel schedule representation and a comprehensive evaluation framework. We evaluate a range of schedule encoding and deep model architecture combinations. The evaluation shows our approach can rapidly generate large, diverse, novel, and realistic synthetic samples of activity schedules.