Keywords: Space Transformed visualization, Parallel coordinates, Andrews plot, Quantization, Binning, Enveloping, Big Data.
Visualizing big data is cumbersome task that require integration multiple techniques from statistics, computer graphics and mathematics in order to deal with visual clutter. In this talk we will cover several techniques such as binning, density estimation, quantization and enveloping to scale our visualization techniques to large number of observations and detect hidden patterns in timely fashion. We focused on space transformed visualization (parallel coordinates and Andrews plot) to represent high dimensional data in 2D space and detect the frequency of each profile in lower space. The frequency is used to saturate similar profile and patterns. We also discuss effective enveloping method to show the pattern via its statistical summary, which in terns enables visualization of multiple classes and understand the underlying geometries regardless the size of each class or the entire data. Several real examples will be presented in this talk.