Interactive and Dynamic Visual Overviews of Large Multi-Dimensional Data

The proposed research seeks to develop sophisticated dynamic graphics tools interactive exploratory analysis of very large datasets. These tools, when used in conjunction with data mining algorithms, will enable the user to (1) detect local departures from an overall trend, (2) detect structure in sparse high dimensional data, (3) select data dependent preprocessing steps (e.g., scaling., normalization, feature extraction,) to enhance the effectiveness of automated numeric or symbolic 'machine learning' algorithms, (4) refining solutions obtained with machine learning algorithms, (5) explore the differences in inductive and representational biases of alternative machine learning algorithms (e.g., decision trees versus neural networks), and (6) overview the data space as well as the complex relationships discovered by the data mining algorithms.