Traffic data mining project

Data Set description

• Sytadin: travel time and instantaneous speed measured on fixed stations detectors (magnetic loops) installed all over Paris agglomeration highway network. 1606 stations with frequency of one minute during 3 months.(data here...)

Main possible goals

• (i) Data analysis: search for spatial, temporal and spatial+temporal features
• (ii) Traffic reconstruction problem: predict the data corresponding to silent stations given observations (speed, travel time, density..) from active ones for say 5%, 10%, 20%, 50% and 80% of randomly selected active stations.
• (iii) Traffic prediction problem: given available observations (speed, travel time, density..) at time t, predict states at time t+h for all stations, for e.g. time horizon h=10, 20, 30, 60 minutes.
• (iv) Open problem: find something meaningful into the data.
• (v) ...

Contact: Cyril Furtlehner cyril.furtlehner@inria.fr