8570_en-US

liliconcom

Scatterplot Matrix nonpar density<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />

Hello,

I've looked through a number of resources trying to answer this question, but I've been unccessful in finding/understanding the answers available.

After I create a scatterplot matrix, I select the nonpar density option in the Scatterplot Matrix Menu (red down arrow). This creates a red shaded zone in the densest section of the graphs as well as a gray area that surrounds it. Obviously the gray area is less dense than the red area, but how do I interpret what these shades represent?

The only answer I have found so far is this....

Shows or hides the nonparametric density, which represents the areas where the data points are the most dense. The nonparametric density estimation is helpful when you have a lot of points and the density of the points is difficult to see. There are two quantile density contours. One contour includes 50% of the points, and the other contour includes 100% of the points.

To me, this means that the red should be the densest area of the scatterplot encompassing 50 percent of the points while the gray area should encompass everything else. The problem with that is that the gray area does not encompass the rest of the data.

Any help or direction will be greatly appreciated. Thanks!

Re: Scatterplot Matrix nonpar density

I think the doc is wrong and the outer density contour actually represents 90% of the data. You can compare it to the nonpar density in Fit Y by X (uncheck 5% to see 10% intervals).

I'll need to confirm that, but regardless of the actual number, it will not represent that many points exactly because it's really 50% or 90% of the smoothed density, which may not map perfectly to the data.