Which FIT files should I include for calibration?
More files are not always better.
The best calibration dataset is usually the one that best represents how you actually move on the trips you want to plan.
Include files that are:
- representative of hiking or backpacking you actually do
- spread across meaningful terrain variation
- typical of your real pacing and effort
- believable enough that you would trust them as examples of your normal movement
Exclude files that are:
- from unrelated sports
- obviously corrupted or incomplete
- dominated by stop time or strange recording behavior
- extreme one-offs that do not represent how you usually travel
Decision rule
Ask:
If TRIPS learned from this activity, would that make future backpacking plans more realistic or less realistic?
If the answer is less realistic, leave it out.
Start smaller if you are unsure
If you are torn between a smaller clean set and a larger messy set, start with the smaller clean set.
You can always expand later.
Why this matters
Calibration is about improving relevance, not maximizing file count.
A moderate but representative dataset is often more useful than a large mixed dataset.