In the Vehicle Activity Events list, the Type column is a dropdown. Click it and pick what actually happened:
- Delivery
- Pushup
- Clean out
The change saves immediately — Event recategorized as Delivery. Clean-out rows are shown in amber so they stand out.
Undoing it
Once you have overridden an event, a ↩ button appears to revert it. Click it and you get Reverted to ML prediction. — back to what the system originally decided.
Clean out
This one is worth knowing about even if you never use the others.
A bunk clean-out — scraping out refusals before fresh feed goes in — looks a lot like a feed delivery to a camera. Machine arrives, works along the bunk, bunk contents change. Left alone, a nightly clean-out can show up as an extra daily delivery, which inflates your delivery count and skews your feeding-time consistency.
Tagging it as Clean out takes it out of both the delivery and push-up numbers, and it stops distorting your schedule grading, while still keeping the event on the record.
If your Avg Per Day deliveries reads one higher than you know you feed, and there is an event at the same time every night, this is almost certainly why.
Re-tag or reject?
Re-tag whenever the event genuinely happened. Rejecting a mislabelled push-up removes it from the delivery count and the push-up count, so you have swapped one wrong number for two.
Reject only when the event should not be counted at all — see Accept or reject an event.
Refresh afterwards
Like rejections, re-tagging makes the summary numbers stale. Click Refresh on the Statistics outdated card when you have finished making changes.
The faster route for a repeated mistake
If the same machine is misread over and over — every clean-out counted as a delivery, every push-up missed — do not fix them one by one. Tag your feed vehicles and the whole batch is re-checked with your answer.
Use re-tagging for one-offs; use vehicle tagging for patterns.