This week’s project at work is to catch and reject any data that has location coordinates that are too far away from the training range. In and of itself, detecting an out of range coordinate is fairly trivial (especially since we’re using UTM coordinates rather than Lat/Lon), its the permutations of the type and shape of the obstacles that make this complicated. So so far I’ve come up with about 9-12 different convoluted involved cases I need to test to make sure this works “properly”…double that if I want to regression test the non-error condition cases. But the good news is, I’m making forward progress and I’m trying to approach this methodically so I don’t miss any major cases.