Exploring Big Data Landscapes with a Glyph-based Zoomable User Interface

Abstract

High-dimensional data sets are hard to explore using common spreadsheet environments. However, data scientists need to develop appropriate clustering and classification algorithms to make sense of big data repositories. Even sophisticated analysis tools often focus on mathematical tasks and offer only basic data visualization with few interactive features. In order to gain more sophisticated insights and test hypotheses with regards to high-dimensional data sets, we developed an interactive zoomable user interface using glyph-based visualizations. The visualization is based on a two-dimensional plot of the data space using multi-dimensional reduction. The resulting Big Data Landscapes are then explored with various controls, filters, and details on demand.