Daniel Keim

Interactive Visualization and Feature Transformation for Multidimensional Data Projection

Projecfing multidimensional data to a lower-dimensional visual displayas a scatter-plot-llke visualization is a common approach for analyzing mullidimensional data. Many dimension reduclion techniques existfor perxad forming such a tasle, but the quallty of projections varies in terms of both preserving the original data structure and avoiding cluttered visual displays. In this papel; we propose an interactive feature transformation approach that allows the analyst to monitor and improve the projection quality by transfonning feature space and assessxad ing/comparing the quality of different projection results. The method integrates feature selection and transforxad mation as weil as a variety of projection quality measures to help analyst generale uncluttered projections that preserve the struclural properties of the data. These projections enhance the visual analysis process and provide a better understanding of data.