VIPR: An Interactive Tool for Meaningful Visualization of High-Dimensional Data / 4274
Donghan Wang, Madalina Fiterau, Artur Dubrawski
Analysis, pattern discovery, and decision support can all benefit greatly from informative and interpretable visualizations, especially of high-dimensional data. Informative Projection Ensemble (IPE) methodology has proven effective in finding interpretable renderings of high-dimensional data that reveal hidden low-dimensional structures in data if such structures exist. In this demonstration, we present a powerful analysis tool that uses IPE methodology in support of fundamental machine learning tasks: regression, classification, and clustering. Our tool is an interactive web application operating on 2D and 3D projections of data automatically selected by IPE algorithms as informative for the user-specified data and task.It also provides RESTful APIs enabling remote users to seamlessly integrate our service with other tools and to easily extend its functionality. We show in examples how it can discover hidden interpretable structures embedded in high-dimensional data.