Province of North-Brabant
Analysing, visualizing and predicting biking behavior
The Netherlands is the worlds biggest cycling country. No less than 36% of the dutch sees the bike as the most important mode of transport in daily life. Cycling is also the most sustainable mode of transport, making it something that should be encouraged. An appropriate infrastructure which keeps the desires of cyclists into account is therefore of high importance.
Province of Noord Brabant
The province is responsible for infrastructure development in Noord-Brabant; including cycling. Based on available information, decisions are being made regarding development of cycling paths, enhancing safety on crossroads, and optimizing traffic flow.
The amount of information available to decision makers is generally limited. A lot of data is available on secondary topics such as accidents on road works, but a complete overview of how many cyclists cycle where at a certain location at a certain moment does not exist. Various initiatives have attempted data collection on cycling behavior, yet these lack comprehensiveness. On the one hand, there are mobile apps which people can install on their phones, which collect data on how these people use their bikes. The problem is that only a very small proportion op the population uses such apps, making that the sample is likely not a fair representation of the entire population. On the other hand, we have bike-counting-points, which are cables spanned over cycling paths which count how many cyclists pass by. This solution can collect data of the whole population at every moment, but it can only be installed at a limited number of locations. There is a need for a comprehensive overview of how all people cycle through cities, all the time and everywhere.
The Solution: Machine Learning
In collaboration with the Province of Noord-Brabant and Breda University of Applied Sciences, we work on a solution to provide more insight into biking behavior.
At first, we developed a visualization platform in which current data sets are presented, as can be seen on the right. This provides insight into what data is available, and clarifies what the focus areas can be for creating a better solution.
Following this, we develop a machine learning model which predicts how many cyclists will be at a certain place and time based on the available data. To make this possible, we combine data collected through all aforementioned methods, and enrich that by for example data on traffic or the weather. We investigate the predictive value of these variables on how many cyclists will be at a certain location under specific circumstances. This part of the solution is currently still under development.