Midday runs may look similar in the middle but differ at the extremes. This distribution does not prove midday causes deviation. Verify identities and time basis.
Comparing the tail
For each band, show run count, middle and highest value. Their spread exposes unusual runs; one number can hide sample count.
Verifying time fields
Separate local time, logger clock and download time. A mismatch can put records in the wrong band. Keep the original file and document alignment.
Sample coverage
State each band's valid record count. A missing measurement is not normal temperature; exclude gaps from extremes. Few runs do not show a recurring pattern.
Middle and extreme values
For a standout run, match start and end times, route stage, delivery and file integrity. This adds context; correlation alone proves no cause.
What to compare. Group by route, similar load and measurement definition. Do not combine records after a definition change. Document field trials under the authorized quality process.
Season and route effects
For monthly comparisons, retain season, route and load; otherwise band-specific explanations are unreliable. Repeat with comparable records.
Limiting the conclusion. Name the runs behind a longer midday tail. This selects examples for review, not cause or product suitability; follow specifications and quality review.
