City Dashboard: A framework for spatiotemporal analytics of transportation data
Frédéric Roulland, Michael Niemaz, Luis Ulloa, Christelle Loiodice, Jeffrey Kingsley
Transportation is one domain that has embraced the big data era. There is a fast growing presence in cities of sensors connected to a centralized system (traffic cameras, parking management systems, toll gates, fleet management systems ...). Transportation authorities and operators aggregate a massive amount of operational data. Monitoring the city mobility patterns and being able to plan the development of the infrastructure according to this data is a grail quest that many want to pursue.
However extracting value out of this massive and heterogeneous data creates some challenges which still deserve some research. Without considering all the technical, organisational and political issues associated with the integration of these various sources of data and assuming these data are made available getting value out of it is still not trivial. Since mobility planning for a city area has a political dimension, visual analytics approaches, which offer the possibility to understand a situation and let humans in control of the decision are often more adapted than automated decision support approaches.
Although visual analytics on big data will be a key element of next generation technologies in the transportation domain, to our knowledge, today very few solutions enable decision makers an easy and direct monitoring and access to their operational data. On one hand, operational solutions such as parking management systems or fleet management systems have limited visual analytics capabilities and are always bounded to the specific data they monitor. On the other hand, multimodal studies of mobility patterns of a city area are mostly performed by consultancy experts. They cannot be requested on demand in daily routines and they are often very expensive because they require a lot of manual aggregation of the various sources of data. We have designed a framework for addressing this gap and developed a prototype platform based on this framework that we call City Dashboard.
The City Dashboard framework enables building a single visual analytics platform on top of the various transportation operations data of a city area. One core principle enabling the flexibility of the platform is a complete decoupling between the processing of the business meaning of the data and the way to represent it. Data are represented through Visualisations which consider only the spatiotemporal attributes and the nature (discrete quantity, distribution ...) of each feature. Services process the data and give a meaningful result for a particular domain. Using a wizard both Service and Visualisations respectively implemented by a domain expert analyst or by a data visualisation expert can be easily coupled in the platform. For a given city, the City Dashboard will first dynamically list all the services available. Once a service is selected and provided with some arguments, a list of visualisations compatible with the output of the service is suggested. The services and visualizations are controlled from a web-based GIS client and layered on top of a map of the city area which enables an easy and immediate analysis of any of the transportation data. Therefore the final view is one of the combinations of services-results with compatible-visualizations with layer-orders. Thanks to this approach we have developed a prototype platform that comprehends more than twenty services and fifteen visualisations to analyze data from on-street and off-street parking, Public transport fare collection and fleet management of four cities in Europe and US. In our talk we will be able to illustrate the principles of the framework with a live, real-data demonstration of our platform.
The 7th International Visualization in Transportation Symposium, Irvine, California, October 23-25, 2013.