Patterns and Outlier Analysis of Traffic Flow Using Data Signatures via IDIRBrG Method and Vector Fusion Visualization

• Published in: 2010 3rd International Conference on Human-Centric Computing pp. 1 - 6
• Main Authors: Malinao, Jasmine A, Juayong, Richelle Ann B, Becerral, Jona G, Cabreros, Karmina Rosette C, Remaneses, Ken Marvi B, Khaw, Jason G, Wuysang, Danny F, Corpuz, Francis James O, Hernandez, Nestine Hope S, Yap, Jan Michael C, Adorna, Henry N
• Format: Conference Proceeding
• Language: English; Japanese
• Published: IEEE 01-08-2010
• Subjects: Clustering algorithms; Data analysis; Data models; Data visualization; Discrete Fourier transforms; Time domain analysis; Visualization
• ISBN: 1424475678; 9781424475674
• DOI: 10.1109/HUMANCOM.2010.5563344

In this paper, we used X-Means clustering algorithm, incorporated data images from a so-called Iterative Data Image Rotated Bar Graph (iDIRBrG) method (formerly referred as BC method) and used Vector Fusion Visualization to achieve better traffic data analysis results compared to our previous work by improving how data signatures are constructed from the raw data set. By doing so, we effectively identify more patterns, extracted known as well as unexpected outliers, and novel information from the data set. These results were validated by experts from the NCTS. Furthermore, we also show more simplistic frequency-domain 2D visualizations of the entire data set. These successfully expose the internal structure and relationship of each individual point as well as the clusters obtained from it. Hence, using the methods above, we provide an effective and efficient time series traffic analysis.