The following can be used for both pedagogical and exploratory purposes. The simple idea is that we have a multi-way table that was created with binned continuous variables and discrete variables. For every pair of variables in such a data frame, Xi and Xj, we have a m x n 2 way table. When considered as a complete data set, we have a hypercube. The intent of the GUI is to allow the user to modify the bins (for continuous variables) interactively and to provide a simple mechanism for exploring the structure in the whole table using histograms, scatter plots and different statistics such as a the Chi-Squared or some variant of that.

A given 2 way table is displayed in the main window. (The user can create other windows or panels in the main window to have simultaneous views of different possible 2 way tables.) The frequencies in the cells of the table are displayed and the margins are shown in the usual positions. Labels for the bins for discrete variables are centered in the margin of the row or column. However, continuous variables are displayed with a scale that runs along the side of the table from the minimum to the maximum value of that variable. The bins are indicated with lines which can be dragged uni-directionally to specify different bins. As the bins change, the cell entries change and so the statistic(s) being displayed

There is also a set of 4 arrows pointing in the directions of the compass. These allow the user to spin the cube to display the tables for other pairs of variables in the data set.

In addition to the usual Help facilities, the menus offer different views of the data including cloned windows so that multiple tables can be viewed simultaneously, histograms and scatter plots, etc. Also, other datasets can be viewed in any window which brakes the link between that window and the window it was originally cloned from.


Last modified: Sun Feb 16 12:48:33 1997