I am using a matcher transformer to match multiple shapes. I have two shapes that are almost a match but are off by .001 of a decimal. In the parameters, I have setting the Vector Tolerance to 1, .01, .001, .00001 and each of those make them a match. We are unsure of how much vector tolerance we should allow because we want these shapes to be as close of a match as possible.
Page 1 / 1
This is probably not the answer you're looking for but...
It depends...
It depends on your data and the units of your coordinate system. From the transformer help:
“Vector Tolerance” allows two geometries to have spatial variation and still be considered a match. It only applies to vector based geometries (points, curves, areas). It is similar to a Fréchet distance: You have a dog leash of length “vector tolerance.” You will walk along the boundary of one of the geometries, your dog will walk along the boundary of the other. If you can both complete your walk without dropping the leash, the two geometries are within tolerance. This transformer has the added constraint for curves that you and your dog must both start at one end of your curve, and end at the other.
So really you are the only one who can determine what an appropriate value would be. If there are, for example, locations of lamp posts in order to send out a maintenance crew I'd accept a tolerance of a few cm (in the same order of magnitude as the diameter of the lamp post), if they're parcel boundaries I'd want it to be more accurate.
I hope this helps.