This study aims to identify mid- and long-term characteristic congestion trends in the urban area by classifying time-series data collected at sensor-installed points using the k-means method a major unsupervised clustering technique, and to support measure planning for each point using the results obtained from the classification. In this study, temporally and spatially characteristic congestion patterns were extracted from a large amount of congestion data obtained from sensors installed at approximately 2,200 locations across Sapporo urban area. The identification of regular congestion patterns that occur at certain locations and hours is expected to facilitate support for the planning of traffic measures that require temporal and spatial consideration. As the result of this study, congestion trends and congestion-point distributions in the city were then classified into a number of patterns, allowing the selection of effective measures and the identification of targets for countermeasures.
展开▼
