Plot selection over time should produce quartile area plots
This issue was created automatically from an original Mantis Issue. Further discussion may take place here.
When you do a plot selection over time, the resulting series are actually descriptive statistics over each field (min, max, average, quartiles, stddev, etc.). This is because a query often returns multiple values, so the descriptive statistics summarize everything to a single value.
That makes sense, but the result of running the filter is this plot that blindly charts all these independent statistical values for all these independent fields, which usually results in an unreadable mess (such as reported in bug #14940 (closed)). We need a better way to present this data.
I propose presenting this data in a quartile area plot. I think I made this name up (I don't know if they have a name), but the idea is that the average value is shown as a deeply colored line, the 2nd/3rd quartile is drawn under that as a muted area, and the min/max range is drawn under that as a more muted area. I have attached an example image of such a plot for one series.
This is a much cleaner and more natural representation of the statistical data. It also has the great advantage where if the selection is returning one item per time step, the areas collapse to show that single deep line. That is, you are getting the exact line you were expecting in the first place.