I have about a million data points now. Some are likely of low quality; I haven't worried too much about that yet. I'm not logging altitude or heading, so some of the various "degraded" GPS modes might work for me.
The next step is to do something more useful with the normalized data. First I need to fill in some dead spaces; for example, if the GPS dies while I'm at work and I don't go out for another day, that would put all the accumulated time into the wrong location. The current best guess algorithm for that is to scan for large breaks in the time sequences that have measurable changes in distance, then prompt for human intervention with the details. If the last entry is arriving at work on a Tuesday morning, and the next is leaving home Wednesday morning, I can at least guess how long I usually remain at work and when I got home. I'll lose the transit points, which I consider inconsequential for this project.
I may not like them, but I'll have them.