Navigating Large Cubes
Once a cube reaches a size of about 100,000 records, the display time for plots starts to become longer than 2 or 3 seconds (for newer computers). This is still tolerable for navigating across different SVs (SyncrhoViews) and windows while selecting this or that parameter (which refresh the plots), but beyond this it becomes increasingly frustrating.
Antaeus provides the following mechanisms for reducing the time involved when working with large cubes.
Pixel plot symbols. This can be an effective way to display dense plots. Pixels draw about 4 times faster than circle plot symbols, and the effects can be striking. Here is a pixel plot containing 120,000 data points:
Sunflower Plots. For extremely large cubes, you can use sunflower plots instead of scatter plots to get some idea of the data until you are ready to look at the final scatter plot. Representing a scatter plot by a sunflower plot is analogous to representing a quantile plot by a histogram. They refresh very fast because counting the data points for the sunflowers is perhaps 20 times faster than drawing them. So a ten million record sunflower plot will draw almost as fast as one of a million records since each would contain approximately the same number of sunflowers. See The Sunflower Plot for a more detailed explanation.
Sampling. You can turn sampling on or off at any time. When you turn it on, you select a sample size from a list of up to10 choices, ranging from 100 to 2 million records, which are randomly selected. The available choices depend on the total number of records in the cube. It is surprising how closely the distribution of the sample plot resembles that of the full plot even for small samples. The following plot is a sample of 10,000 records selected from a cube with 2.5 million records:
Sampling is an extremely effective way of navigating very large cubes. A single click turns sampling off when you're done setting parameters and looking at various plots, letting you save those longer waits for only those plots of most interest.
Escape Key. You can abort the drawing of any plot by simply pressing the escape key. This instantly stops the process and you can go on with your navigation. This is especially valuable for aborting the refresh of a scatter plot matrix or array after the first few cells have been drawn of, say, 36 cells. The partially drawn plot can still be used, however. For example, in a single or double scatter plot, any displayed data point can be clicked on to bring up the record contents window for that data point.