The art of transport modelling: a quantum shift

What key changes have taken place in transport modelling over the past decade, and what might the future hold? Ahead of Modelling World 2015 on June 4, Systra’s modelling experts share their thoughts

The art of modelling has never been easy.  But now, thanks to new technology and methodologies, benefits have increased dramatically and are creating a range of new opportunities.   Data compilation is now much easier, both sourcing and production, with vastly improved outputs.

Systra's Director of Innovation, David Connolly says: “We run our data overnight, but now when we come back in the morning we end up with 10 times what we used to just a few years ago."  

But although computer technology has proved a real game changer in terms of data analysis, the trick is ensuring that the data is as representative of current transport conditions in the first place.

John Allan, Systra’s Director of Northern England, believes that a future game changer is data collection from mobile phone providers. 'We’ve been involved in several projects using data from mobile phone providers, and the early signs show that results are good, but we’ve really got to interrogate this data capture technique to fully understand its capabilities as well as its flaws,' he said.

The practice of collecting information from smartphones is known as ‘data mining'.  The greatest advantage is the size of the dataset, which enables people movements to be pinpointed to specific locations as the mobile phone user travels along a route, in effect mapping an actual graph showing real traffic trends. This data, if used correctly, can help modellers reduce error margins in defining base matrices resulting in models that better reflect journeys within an area.

Yet any use of mobile phone data is a highly controversial subject as it relates to the wider debate about privacy and individual freedom.  David Connolly, however, believes that much of the concern can be overcome: "From a data protection point of view, as long as nobody tries to connect a given phone to a specific individual, there is no problem as the data is simply pooled anonymously. The data is then assembled into wider zones, giving us a very accurate picture of movements within that defined area."

Smartphone and tablet technology now provides travellers with far more accurate real-time information on public transport or highway network congestion that was ever available in the past. This has resulted in changing attitudes and a modal shift from car to bus and train, especially on ad-hoc journeys.  Such behavioural patterns are difficult to predict and pose a headache for modellers trying to keep ahead of the game. 

“None of the current models take account of this change of behaviour, particularly for the infrequent journey,” says David Connolly. “A combination of the internet, planning apps, better pricing and apps that show you when the bus is due to arrive are all having a major effect on people’s perception of public transport,” he added. 

The challenge facing modellers is two-fold.  Firstly, predicting the amount of information received and quantifying the availability of that real-time information. Secondly, even if you can quantify it, how do modellers turn that prediction into behavioural change?  David Connolly believes the answer is the ‘old-fashioned’ survey. He said: “Firstly do a simple poll to find out how many people have the relevant app, and then ask whether this has influenced their travelling patterns.” He added that once you have access to this information, there is less of a requirement to model at an early stage as modellers have such a good data source at their fingertips. 

Another area of major change especially in cities is the rise in cycling and walking. Systra’s London-based management consultant Nicola Troll believes that, given evolving trends, we need to question what exactly it is that we need to be modelling. “One of the issues being debated is whether forecast assumptions and factors that predict behavioural change,  and that have not changed much in past years, are still valid,” she said. “The recent recession has made modelling particularly challenging and we can be blinded by too much data." One area of debate is whether there should be a focus on cheaper, high level, strategic ‘first principle’ based solutions rather than simply developing complex and expensive ‘one size fits all’ models, with functionality that is rarely used or even required.

While technology has dramatically shifted travel patterns, it also gives modellers an advantage in terms of how they map those changes.  John Allan recalls how difficult and time-consuming it used to be physically visiting each location. “Now we can visit a site from our desktop using Google maps, whereas before you could spend a whole day just checking one small road junction,” he said.  “This has allowed us to really focus resources efficiently.”

Systra’s modelling team agrees that, as technology and travel patterns change, the need for modelling techniques to reflect those changes has never been greater. David Connelly sums up: “Tighter fiscal control will have a significant impact on what is modelled and how it is modelled. Governments demand robust appraisal at far lower cost, and as an industry we need to meet this challenge. New technology and the innovative nature of our industry will make sure we deliver."