Industry field guide
Peanut lot and inspection fraction data for AI
A farmer-stock peanut sample may be described at more than one stage.
- 2022 NAICS
- 111992
- Practice pack
- Example and answer key (JSON)
- Updated
What this guide tests
What fraction follows for S1 under the explicit exercise definition, and can S11's fraction be computed or substituted?
See the specimen and fields · Work through the fictional records
Task and source context
A farmer-stock peanut sample may be described at more than one stage. A percentage for a separated fraction is easy to attach to the wrong denominator. USDA’s farmer-stock peanut inspection instructions show a real-world workflow for tracking sample components. A grower can design a records benchmark using invented measurements and an independently written rubric. USDA farmer-stock peanut inspection instructions
Give the model a sample packet with whole pods, separated kernels, and a removed material record. State which weights refer to the original portion and which refer to a later stage. Add a result from a neighboring sample with a similar ticket number. Ask for the correct association and which denominator is relevant for a named calculation. One intentionally incomplete example should reward asking for the original weight.
Keep buyer grading and settlement reports, historic discounts, and contract information private. Use owned photo props and fictional IDs. Keep any food-safety determination outside the benchmark.
Industry research
The U.S. Census Bureau’s industry definitions establish which activities these codes cover. The workflow sources in this guide describe the task or record type. Fiund’s pilot below uses fictional material to test a specific decision within that scope.
111992 · Peanut Farming
This U.S. industry comprises establishments primarily engaged in growing peanuts.
Build a pilot sample
Set up this pilot
Use dummy sample mass tickets before and after separation. Mark the named denominator and keep another sample's missing fraction outside the calculation.
Prepare the test packets
- Start with the worked case. Use the fictional record types below as a guide to the source files you need. Give each packet a stable ID and save its source versions with it.
- Change one fact. Make one packet with enough evidence to answer the question and one where the evidence points to a different answer. Keep the other fields fixed so a reviewer can tell what changed.
- Remove the decisive evidence. Give the reviewer a third packet with a plausible near-match in its place. Have them write what can still be established and what evidence would resolve the rest.
Fields the reviewer needs
- Sample ID and measurement stage
- Connects each component correctly.
- Mass unit and named denominator
- Makes the arithmetic reproducible.
- Neighbor-ticket identity and missingness
- Prevents cross-sample substitution.
Use Plan a farm, forest or fishery AI data sample to turn the specimen into a brief with a defined task, source and review owner.
Fictional exercise
A kernel fraction needs its named denominator and exact sample identity
Invented peanut sample records use practice mass units and an explicitly defined arithmetic task. They do not reproduce an official grading calculation or determine food safety.
| Record ID | Source | What it records |
|---|---|---|
| PF1 | Sample S1 stages | Original test portion 100 mass units. Removed material 10; remaining portion 90. Separated kernel fraction 54. |
| PF2 | Practice question definition | For this exercise only, report kernel fraction divided by the original test portion, not by the post-removal portion. |
| PF3 | Neighbor sample S11 | Kernel fraction 60, original portion missing. Ticket format resembles S1 but sample ID differs. |
Question for the model. What fraction follows for S1 under the explicit exercise definition, and can S11's fraction be computed or substituted?
Show the reference answer
PF1 and PF2 define S1's calculation as 54 divided by the original 100, or 54%. Dividing by the 90-unit remainder would yield 60% but answer a different denominator question. PF3 belongs to S11 and lacks its original portion, so its defined fraction cannot be computed or substituted into S1. Preserve stage, sample identity and denominator rather than accepting a plausible-looking percentage from a nearby ticket.
How to check the result
Ask for the answer, the record IDs that support it and any fact the evidence leaves unresolved. Have a reviewer compare the response against the reference before deciding whether the sample is ready to expand.
Errors to flag in this case
- Using the post-removal mass when the task explicitly names the original portion.
- Borrowing a missing sample weight or kernel value from a similarly numbered ticket.
Change the inputs and run it again
Change the question to the post-removal denominator while retaining all records. Require 60% with an explicit basis rather than memorized 54%.
Score the decision and the evidence separately
Save the reviewer’s expected answer, supporting record IDs and unresolved fields in a separate answer key. Score the model on the decision, its citations and its handling of the deliberate gap. A correct answer citing a different case counts as an evidence error.
Before expanding collection, have a second reviewer run the same checks on the held-out packet. Keep related versions, views and exports of the same case together so they stay on one side of the training and evaluation split.
Check permission and scope
Use original dummy tickets and owned props. Exclude settlement reports, discounts and processor tables; the arithmetic is a fictional evaluation rule.
Remove precise farm coordinates and operator identifiers from a review copy. Keep the linkage between lot, sample and outcome intact.
Describe an authorized archive through Fiund’s existing-data assessment, or describe a new capture capability through the collection brief. The contributor agreement covers the terms for participation.
Sources and further reading
Resources for this workflow
Company and program reviews
Use these reviews to compare sourcing, capture or evaluation approaches. Confirm the applicable program, accepted sources and terms with the provider.
Compare the approach
truelabel review for teams collecting physical AI data
Compare how a capture team specifies devices, processing and quality review for an outdoor recording pilot.
Compare the approach
Graviti review for workplace video and multimodal training data
Compare a catalog with commissioned multimodal collection when a particular crop, setting or viewpoint matters.
Related industry examples
Industry guide · NAICS 111991
Sugar beet load and sample data for AI reconciliation
A sugar beet load can carry soil and other material along with the crop.
Industry guide · NAICS 111998
Hop variety and lot trace records for AI
An acreage or harvest report can name a variety while a sample tag identifies a particular lot.
Industry guide · NAICS 111910
Tobacco crop lot and revision records for AI
A crop lot can be split after harvest while a later report still shows the original total.