Modeling Traffic Data - Datacity

Modeling Traffic Data

Home > Challenges > Modeling Traffic Data



How can we improve the knowledge of traffic conditions by reducing investment and maintenance costs of traffic sensors?

The Ile de France Highway Authority (the DIRIF) has sensors (electromagnetic loops) installed throughout their highway networks. The data is generated from traffic counters (flow) and average road speed counters. The maintenance of these devices is very expensive and complex as a result of construction works, road closings, and security.

Such constraints often originate from incomplete or unworkable data, further preventing the highway authority from actions that might improve traffic conditions (providing motorists with real-time data, managing the effects of roadwork or public events, regulating access to highways, installing bus lanes, etc.).


The digital revolution has encouraged us to consider increasing the availability of new types of data (both complementary and alternative), such as data produced by GPS software.




  • Use case & expected benefits

  • Available ressources

  • Business Opportunities for Partners

  • Startup Laureate

Use case

The Île de France Highway Authority (DIRIF) is responsible for all of the roads and freeways in the region. The annual maintenance cost of road sensors is millions of Euros. Furthermore, there is limited time to carry out such maintenance works. Renewing every single sensor (2,500) would cost an estimated 42 million Euros. The cost is not sustainable, which forces the DIRIF to optimize their numbers by relying on alternative data sources, all while trying to ensure improved services.


GPS software data provides information on vehicle speeds, but it doesn’t account for the total amount of vehicles on the road. Consequently, this doesn’t provide enough information to accurately observe traffic conditions.


GPS data must thus be coupled with other types of data sensors in order to recreate a reliable understanding of the amount of traffic and vehicle speeds. This must be done without requiring more sensors.

Expected Benefits

This project’s objective is to obtain more reliable traffic condition data and reduce the maintenance costs of the existing infrastructure.


Every single road maintenance authority has a fleet of sensors to maintain, which makes each of them a potential client.


The DIRIF has a wide range of speed counter and traffic counter databases across its network, which can be installed in and around the proposed area of testing:


  • Since 2013, 2500 data loops across the network have been registering data every 6 minutes
  • Since October 2016, sensors have been monitoring average road speed data across the entire network (2500km)


DIRIF: Romain Rémésy, data expert

Setec: Perrine Cazes


  • Coordinator: Ghislain Bourgin
  • Deep learning expert / traffic counter video software: Enzo Fenoglio
  • ‘Sensor fusion’ expert (source data correlation: Wi-fi, Bluetooth, video, etc.): Gaëtan Feige
  • Datalake Expert Franck Bachet

Experimentation field

A suitable interconnected network: a section of highway where there is a sufficient quantity and quality of DIRIF equipment and available data, as well as sufficient data on average road speed.


For example, here are a few possible highway junctions:


  • the RN118 between the A86 and the RN104;
  • the RN104 between the A6 and the A10;
  • the A6 between the RN104 and the A86;
  • the A86 between the RN118 and the A6;
  • the A10 between the A86 and the RN104.

For the start-up, the objective is to market a tool capable of reconstructing missing traffic flow data in conjunction with GPS software data and other speed and traffic counter information.


For Setec, an engineering and consulting agency that also works with highway infrastructure engineers, the developed pilot could help offer new solutions for their clients.


For Cisco, the experiment will provide specific and practical cases for the open-source data analytics platform PNDA.


Wintics is developing tailor-mode algorithmic solutions (Data Science and Operational Research) to address business issues of its clients.  The startup’s team is combining both technical and business expertises to guarantee the relevance of its solutions.

Partners' Experts

Perrine Cazes

Mobility Engineer

Romain Rémésy

Chief of the "Observatory and Traffic Engineering" Department

Franck Bachet

Chief Technical and Architectural Officer

Enzo Fenoglio

Data Scientist

Ghislain Bourgin

Program Manager Business Innovation

Gaëtan Feige

Co-innovation Lead