Research
My research asks how forecasts, which are always uncertain, can be turned into operational policies that are profitable, robust and simple enough to implement. It began in evolutionary computation, moved to forecast-informed pricing and routing for online grocery delivery, and now concentrates on scalable forecasting systems deployed in real businesses.
Scalable forecasting for large product ranges where demand is smooth for some items and sparse or intermittent for others; model selection by demand type; making forecast uncertainty usable in ordering decisions.
Decisions about delivery slots, prices and routes when orders arrive over time. Forecast orders are used to estimate the opportunity cost of each booking; profit and unnecessary driving are both considered.
Earlier work from the MSc period on genetic programming and metaheuristics for regression, program synthesis and differential equations.
4 peer-reviewed journal articles and 1 conference paper.
Bibliographic details are taken from the publishers' records. The 2026 Annals of Operations Research article is open access under a CC BY 4.0 licence and can be downloaded here.
Online grocers must decide which delivery slots to offer and at what price while orders are still arriving. This paper estimates the opportunity cost of each slot more accurately by routing forecast orders alongside confirmed ones, and introduces a dynamic slot-combination strategy that uses customers’ flexibility to make routes more efficient. Both ideas work within standard time-window routing systems and are evaluated on real data.
When a customer books a delivery slot, the retailer does not yet know which other orders will arrive. This paper forecasts the orders still to come and includes them in a dynamic routing model, so that slot availability and delivery charges reflect the true cost of each booking. The approach is tested on real-world attended-home-delivery data.
Genetic programming evolves computer programs, but the search is often slow and unfocused. This paper proposes a rule-centred variant that uses an imperialist competitive algorithm to organise the evolutionary search, improving the efficiency of program synthesis.
Adapts the shuffled frog-leaping metaheuristic to evolve symbolic regression models, offering an alternative to classical genetic programming for learning mathematical expressions from data.
Explores Cartesian genetic programming with a crossover operator as a way of finding approximate solutions to elliptic partial differential equations.
These are working papers and are not counted above. Status will be updated as they progress.
Forecasting and decision support under heterogeneous demand
Draws on the Knowledge Transfer Partnership to compare global forecasting models with local models selected by demand type, and to describe how such a system can be run reliably in a small business.
Manuscript complete; preparing for submission
Demand management, pricing and routing for last-mile delivery
Extends the doctoral work on attended home delivery by including the effect of delivery-price reductions in the estimate of opportunity cost.
Under revision (revise and resubmit)
The next programme of work concerns how organisations move from increasingly powerful predictions to reliable decisions. Four strands follow directly from what I have already done:
I am interested in collaboration with researchers in operations, analytics and information systems, and with companies willing to host applied projects. The Knowledge Transfer Partnership model worked well for me and I would gladly repeat it.