Pattern Mining of Road Traffic in Developing Countries using Spatio-Temporal Data

Abstract: A lot of research has been done on city traffic routing mechanisms and congestion analysis using traffic sensors and CCTV cameras in spatio-temporal mining. Developing countries like India face periods of intensive rain for three-four months a year which has a drastic impact on road traffic. In this paper we have worked on creating a model for mining periodic patterns in traffic, which incorporates the fluctuations that occur due to monsoon in three-four months of the year, thereby depicting a holistic picture of traffic analysis in major cities of developing countries. The crucial aspect in Spatio-Temporal Data Mining is investigating temporal and spatial relations simultaneously as individual dimensions. For this reason it is important to choose a clustering algorithm which gives the most optimum performance for multi-dimensional datasets.

Keywords: Periodic patterns, Traffic Analysis, Spatio-temporal data, DB-Scan, Probability Distribution Matrices.