Whenever I run, I carry a GPX device to track my route. Over time, I’ve collected a large set of personal GPX traces. It made me realize that this personal dataset isn’t particularly valuable when I look at individual traces, but it can be made useful once I’ve collected multiple traces over the same city neighborhood. The problem is that I have no idea where OSM ways are missing. This sounds like a great use case for aggregated personal data, because you tend to run the same streets most of the time. The basic idea is very simple: split GPX files into smaller segments, cluster them, and check whether each cluster maps to a known path or street in OSM. Then create an OSM extract for the new paths, along with the original GPX files, so that edits in JOSM are faster. It works! The slow part is making the edits, but at least you are aware of where OSM is incomplete in your neighborhood. The source code is on GitHub