A Quantitative Analysis-Based Algorithm for Optimal Data Signature Construction of Traffic Data Sets

In this paper, a new set of data signatures is derived to obtain better Vector Fusion 2D visualizations of a time series and periodic nD traffic data set as compared with previous work. The latter had used the entire Power Spectrum components for visualization purposes to produce 2D representations of each subset of the data. With the feasibility of obtaining a smaller representation of the data set in obtaining better cluster models compared to using the original n-dimensions, we now explore this feasibility for visualization purposes. We propose an algorithm that determines, in quantitative terms, how good the selected set of signatures represents the nD data set in 2 dimensions. We use the Vector Fusion visualization algorithm in transforming each signature from its n dimensions into 2 dimensions. An improved set of qualitative criterion is drawn to measure the goodness of the 2D data signature-based visual representation of the original nD data set. Finally, we provide empirical testing and discuss the results.