Using big data from road sensors to improve transportation planning

For policy makers, accurate forecasts of future road use are very important. It allows them to plan road capacity, combat congestion and reduce pollution, researcher Ma says. The data needed for those forecasts has traditionally always been collected with surveys that statistical agencies send out to firms in the transport sector. For companies, filling out these surveys is not only time consuming, the results often also lack in precision because not every company reports their trips in the same detail and with the same time intervals.

Loop detectors

The now available road monitoring techniques each produce data with very specific characteristics, Ma says. Loop detectors that are installed in the road surface on many spots generate data from just one point, but they do detect the length of a vehicle. Cameras above the road can be used to read the number plates and track a vehicle between several points, but the system is not available everywhere. The statistical model that Ma developed to model freight transportation now allows all these different types of data to be combined with the more traditional survey data.

Multiple data sources

With Ma’s results, Statistics Netherlands, the organisation that funded her PhD position, can now create a much more detailed statistical model for predicting freight truck demand, both in time and in place. This is because of the sheer amount of data that automatic monitoring generates. Even for one particular city or a short time period, there is enough data left to draw solid conclusions, especially when combined with the existing transport surveys. And these surveys will probably not disappear the researcher says. As valuable as big data from road monitoring is, in the future surveys under road users still be needed to find out about people’s purpose for a trip and how they combine their trips, she concludes.