SCI TV - Visualizing and Interacting with High Dimensional Data

Comprehending data in spaces with dimensions higher than three has been one of main challenges in visualization research. Among many techniques proposed for exploratory visualization of multidimensional data, parallel coordinates scheme, which represents an N-dimensional data tuples as one polyline crossing parallel axes, has been widely applied. In this talk we will discuss our recent efforts on improving the efficiency and effectiveness of parallel coordinates with the introduction of new interaction methods and visual representations, including the integration of point presentation in parallel coordinates and using local clustering operation for data clustering. We will also demonstrate several application cases of high dimension data visualization on various domains, such as climate simulation, traffic and seismic research.