How to investigate deviations in the same hour band

Group monthly deviations by hour band, then verify which run and container each row belongs to. Recurrence alone does not identify a cause.

Extract the time from the data

Convert start times to one shared local-time definition before counting bands. Timezone, logger clock and download timestamp are separate fields; note the conversion used.

A cluster may appear around morning loading, a break or the last delivery across thirty runs. First divide each band's deviation count by the number of runs in that band.

What happens at that hour?

• Compare door-open durations with deviation start times on the same run timeline.
• If available, match parking location and sun exposure by time and place.
• Check delivery and lid-opening records on the same time axis.
• Compare plate-change time with the model record; do not assume capacity was exhausted.

How do you test an explanation?

These matches suggest investigation paths, not proof of cause. Set out the records each hypothesis predicts; for a door-exposure hypothesis, compare actual open durations and readings before the deviation.

Three deviations in one hour during a month are an initial signal. In the next period, match comparable runs and exposures; if records are missing, leave the result uncertain.

An hour band helps locate records; it does not establish root cause by itself.

Hour or run position?

If deviations also begin at the same stop number, time and sequence may be mixed. Compare different stops in the same hour band and runs reaching the same stop at different times.

Records to inspect by hour band