← All cases

AI Empowers · Public sector

Smarter road maintenance for the City of Ghent

Predict which of 1,403 km of roads need resurfacing before they need full reconstruction.

  • Public sector
  • Predictive analytics
Client
SectorPublic sector
CategoryAI Empowers
StatusPilot completed in 2025
3 yrsroad quality forecast ahead

The problem

Ghent maintains 1,403 km of roads. Resurfacing in time is far cheaper than rebuilding, but knowing when is hard.

The solution

An AI opportunity workshop produced three ideas. The first became a predictive model on 20 years of road data, combined with De Lijn, traffic and environmental data.

The outcome

A pilot that forecasts road quality three years ahead, starting with car roads, then bike and foot paths.

21AI use cases from two workshops
40 / 50borderline areas called right four years ahead, against 28 for the baseline
XGBoostthe best of the models we tested

The full story

  1. From ambition to roadmap

    It started with an honest question from the city: "We believe AI can help us, but we're not sure how." Two workshops with city officials and District09 in March 2024 surfaced 21 possible AI use cases across departments. Three were worked out further, and the timing of road resurfacing stood out.

  2. Why timing matters

    Resurface too late and the damage reaches the structure underneath, which means a full and expensive reconstruction. Resurface just in time and a new top layer is enough. The goal: predict that window, road area by road area.

  3. The model

    The city's road inspections were combined with bus counts, traffic models and neighbourhood data. Street segments changed too much over the years, so the model works per area. We compared linear models, decision trees, random forests and gradient boosting: XGBoost won, and on a dataset this size simple models beat complex ones.

  4. Results and next steps

    Four years ahead, the model picked out 40 of 50 borderline areas correctly, against 28 for the baseline. That longer horizon is exactly where road planning needs help. A Power BI tool shows the results to the roads department, and the approach can extend to bike paths and footpaths.

Tech

Random forestGradient boostingLinear modelsNeural networks
This tool allows us to schedule minor interventions earlier and more efficiently. In doing so, we can avoid costly reconstructions and minimize disruption.Joris Vandenbroucke, Alderman (Schepen), City of Ghent
Facing something similar?We would shape your version around your own data.

Ask Pai

Ask Pai about this case

Pai is an AI. It only knows what is written on this page.

Baby Pai · AIHi, I'm Baby Pai, an AI assistant. Ask me about the City of Ghent (District 09) case.

Pai is an AI. Answers come from this case page. Don't share sensitive data.

Chats are kept for 30 days unless you switch that off in Privacy and choices.