Visualisation and Analysis of Urban Traffic , Data

Data on vehicle operating conditions have been recorded by a specially equipped car. while circulating through Naples on several typical routes. during rush hours. Traffic was defined in terms of parts of movements between two consecutive stops ("kinematic sequences"). characterized by a number of kinematic variables. such as mean velocity. driving and idling time. gear ratio. number of peaks. etc. Cluster analysis was used to identify typical groups of sequences. and was followed by canonical discriminative analysis yielding canonical scores used in plots. Series of successive sequences were then grouped into cycles. based on observed transition probabilities among clusters. Cycles were then clustered. yielding seven groups of typical cycles that could be used to characterize traffic intensity and to visualise traffic conditions. Furthermore. the cycles could be used to simulate typical driving conditions and to measure exhaust emissions and gas consumption in the laboratory. Several different visualisation techniques are proposed. such as time and distance plots of the sequences for visualising traffic along different routes. taken at various time of the day. and for identifying highly congested traffic. or highly polluted zones. Other plots of interest include plots of consecutive sequences by their canonical scores. that distinguish between groups of sequences corresponding to different traffic levels. two and three dimensional representation of the groups of sequences and cycles. etc. Also is described how the video images recorded by video camera can be integrated with other traffic visualisation tools.