INDUSTRY FIELD GUIDE
Cattle feedlot pen transition data for AI evaluation
A feedlot’s pen roster is time-dependent.
NAICS 112112 · Cattle Feedlots
A feedlot’s pen roster is time-dependent. A scheduled move and a confirmed move can look similar in an export but should produce different membership. APHIS’s Beef Feedlot study documents practices around animal identification. A feedlot can create a fictional sequence that tests whether a model reconstructs a past roster from the evidence available then. APHIS beef feedlot identification study
Give the model a starting pen roster, an intended move, a confirmed movement entry and a later correction. Ask who was in each pen at a specified time, with reasons. Add an aggregate count that matches only after the correction. Keep a separate key with effective time, event time, and which correction supersedes the earlier entry.
Exclude animal health events, contracts, true inventory counts, and premises details. Use an invented yard layout. Keep recording around live cattle, gates, and equipment outside a first data sample.
Describe a pen records benchmark
Original Fiund exercise
A scheduled move does not change the pen roster before confirmation
An invented yard with two pen aliases and artificial animal IDs forms a time-dependent records exercise. No real layout, cattle handling or operational movement occurs.
| Record | Source | What it records |
|---|---|---|
| FP1 | Initial roster | At 08:00, pen P1 contains A and B; P2 contains C. |
| FP2 | Move records | 09:00 plan says move B from P1 to P2 at 10:00. Confirmation logged 10:20 says effective move 10:05. |
| FP3 | Correction | At 11:00, authorized mock correction changes that confirmation's effective time to 10:15, superseding only its time field. Final aggregate P1=1, P2=2. |
Question for the model
Using the corrected history, where was B at 10:10, and what would an as-known-at-10:30 review have seen?
Show the reference answer and checks
FP3's corrected effective time places B in P1 at 10:10, with the move not effective until 10:15. An as-known-at-10:30 review would instead have FP2's uncorrected 10:05 confirmation because the correction was not logged until 11:00. FP2's original schedule alone would not establish execution. Keep effective event time, record availability and correction history separate; the later aggregate confirms only the final counts, not every earlier timestamp.
- Changing roster membership from an intended move rather than a confirmed event.
- Using a later correction in an earlier information set without distinguishing corrected history from what was known then.
Fields to preserve
- Animal and source/destination pen
- Defines the membership change.
- Effective and logged timestamps
- Supports historical and as-known reconstruction.
- Correction target and field
- Limits supersession to its stated scope.
Test another case
Add a planned return move with no confirmation. Require B's final confirmed location to remain P2.
Use a fictional ledger and yard. Exclude health events, premises, contracts and actual inventories; this sample authorizes no live movement.
Discuss a contribution
Describe the sources you control, the task they could support and what has already been checked. Use the existing-data assessment for an archive or the collection brief for a new recording project. The contributor agreement defines Fiund’s licensing authority and compensation. A guide or fictional exercise does not establish an accepted collection or a sale.