Is it ‘disruptive’ to use Mobile Phone Network Data in transport planning?

tftp welcome a guest blog from industry professional, Philippe Perret MSc MCIHT who is the Associate Director and Head of Analytics at Citilogik Ltd. Philippe delivers a session on 'Mobile data in Transport Planning' in the popular 'Managing Transport Modelling' training course, which runs every year in various UK locations.

Philippe talks frankly about Mobile Phone data and what opportunities it can bring to the Transport Planning industry...

"Having a strong interest in this subject, I have read a lot around it and found plenty of strong and contradictory views on whether Mobile Network Data (MND) constitutes a disruptive technology.

In academic literature ‘disruptive technology’ is defined as one “that displaces an established technology and shakes up an industry or is a ground-breaking product that creates a completely new industry.”

In practice, MND is different from other data sources generally available to transport planners to use as baseline data in transport modelling because it covers the whole UK geography and allows large sampling and relatively long survey durations, ironing out the day-to-day variations often seen in traffic patterns. When put like this, the potential for MND to be a disruptive technology sounds very exciting with little need to consider anything else.

But what about the reality?

In reality, raw mobile phone data is only a database of observed movements, derived from anonymised events, covering the whole territory for a relatively long period of time. Advocates of ‘untreated’ MND sometimes fail to mention the problems of spatial accuracy, mode and purpose definition or statistical reliability around the algorithms used to derive observed trips and the expansion mechanisms used to convert the sample trips made by ‘mobile devices’ to journeys made by the full population of ‘real people’. Indeed, if you view raw mobile network data files with these characteristics such information can be seen to have considerable drawbacks which do not make MND sound as exciting as first thought.

At Citi Logik, we have been working (and continue to innovate) to solve some of these difficulties. Using a series of purpose built algorithms we can provide valuable insights such as mode (typically split by slow modes, road based, rail based) or purpose (typically split by commute, home based other, non-home based). Yet these are based on a series of inferences, rather than true observations and as such they are naturally accompanied by biases and limitations (which at least we know about). Expansion is another hot topic, and once again the Citi Logik team have developed procedures to provide the most representative picture possible of trip making (times, origins, destinations). This said, we are forever trying to understand complex, not always rational, people behaviour through part observations, part algorithms. The data science isn’t perfect, a bit like us!

Where does this leave us?

We have certainly seen that MND is more and more seen by transport consultants as a viable option to reduce or replace Road Side Interviews, as it is less encroaching on traffic, safer, simpler to arrange and it allows the capture of wider area movements. This demonstrates that MND is becoming an agent of change in the transport planning industry. Yet, I believe that we are at a time where the transport planning community and data scientists like ourselves at Citi Logik need to continue to work together to evolve and improve the accuracy of the outcomes of MND analytics. Our goals are really the same, enable transport decisions to be made more confidently, using good quality underlying data and advanced modelling techniques. For this to happen, MND providers like Citi Logik are always looking for innovative transport consultants to work with to test our data with challenging scenarios and providing constructive feedback for us to build upon.

In turn, the Citi Logik Team, as advocates of properly ‘treated’ MND, promise to listen and take the time to understand the concerns transport planners have about the quality of mobile phone data and address them in a transparent way. Together, both the advocates and detractors for the use of MND should recognise that other direct observations from traditional survey methods are valuable in enhancing and validating MND analytic output.

So, is MND a ‘disruptive technology’?

The answer is no, or at least not yet.

Is MND an agent of change that can help rethink and add value to the transport planning process?

Most definitely Yes. This new data type offers wide coverage and a lot of observations which, today, has few rivals but other techniques offer some certainty and precision which MND cannot provide, at least for now.

I believe in working with the strengths and weaknesses of all data sources. Hence, a well-designed survey transport plan making use of various data collection techniques is what is required."

To find out more about Mobile Phone Network Data in transport planning…

Join Philippe on the upcoming ‘Managing Transport Modelling’ training course from tftp. Due to held at WSP office in Birmingham on the 7th June 2017. Only 30 spaces available so make sure you book up quickly.

Find out more and book the ‘Managing Transport Modelling Training Course’ here

Managing Transport Modelling training course - book now

#transportplanningevents #transportplanningtraining #transportmodelling #mobiledataintransportplanning #mobileponedata

Recent Posts
Search By Tags
No tags yet.
Follow us
  • Grey LinkedIn Icon
  • Grey Twitter Icon
  • Grey Facebook Icon