Hyperdimensional Data Analysis and Structural Inference
Edward J. Wegman · 1991
This project focused on exploratory and structural inference for high dimensional data. Exploratory techniques focused mainly on the parallel coordinate technique which was exploited on graphical computing platforms. A MS- DOS package was developed using this and other dynamical graphics techniques. Several refinements of the parallel coordinate technique were made including scintillation, parallel coordinate density plots and grand tours in d- dimensions. The structural inference work is nonparametric in character and focuses on high dimensional density estimation and ridge estimation. These ideas are currently still under development and promising techniques involved random tessellations, the trajectory method for finding ridges and the gradient density plot.