Unsupervised Deep Learning on Spatial-Temporal Traffic Data Using Agglomerative Clustering

Nowadays, road traffic arises from the increased population because of employment in urban areas of all places. This creates to be a challenging aspect, uses certain measures to solve then and thereby the authorized people in the situation. The varying types of services related to traffic streams are pedestrian census, turning movement census, and others. The highlighting features are tabulated based on traffic management techniques. The proposed study on agglomerative clustering on traffic data has been satisfied with its enhanced features. Agglomerative clustering is an extensive model of hierarchical clustering, a bottom-up approach that combines the similarities of samples in clusters. A quite number of methods to find optimal clusters in different ways are briefly discussed to support the clustering. Experiments studied on California-Traffic solution-Data from SWITRS conducted with traffic data prove the optimal number of clusters formed, its validity using the method of clustering accuracy and pasteurization of time series on traffic data are shown in different categories.