This Java application allows a user to probe through a database of 1995 sales figures for a fictitious automobile manufacturer. It shows how response times to user input can be much better with local processing than with remote processing, i.e. using HTML forms.
The base data is 52 weekly sales figures for each of 12 models of vehicle for each if 15 sales regions (just over 9000 data values). Time can be viewed as Quarters, Months, or Weeks. The sales regions are grouped into global zones: North America, Europe, and AsiaPacific. The models are grouped by category: Luxury, Sport, Economy, and Utility, and by line: American-styling, European-styling, and Japanese-styling.
Navigation through the data set is done by clicking on fields in the axis descriptions. Clicking on a named field beside the data area goes deeper into the data. Clicking on colour fields further away from the data changes the choice at the given level and/or pulls away from the data. Clicking on a root (the black rectangles to the left and bottom of the viewing area) always returns to the highest level for that axis, giving totals for the year (the initial state).
A demo script can be run to show some of the possible views, and transitions can be either Smooth or Jump, as the user prefers.