I have sets of time series data which I display as charts in mobile applications.
To make the charts clearer, I simplify the sets by applying Ramer-Douglas-Peucker.
If I apply RDP to a small set with an epsilon that's too high, the set becomes over simplified.
Is there any way to asses the appropriate epsilon for a set?
Should I just use the variation of the algorithm that specifies the desired number of points?
EDIT: I've managed to work out that if I know more or less how many points I want the set to contain after simplification I can roughly calculate:
epsilon := (totalNumberOfPoints / (3 * desiredNumberOfPoints)) * 2
But I'd love to learn other techniques.
