Yield Prediction Engine reference profile.
What will crop yield be for this field or planting, given current growing conditions, inputs, and history?
Overview
Forecasts field-level or block-level yield ahead of harvest, grounded in the operation's own soil, weather, and input history rather than regional averages alone.
Who owns this decision
VP Agronomy, Head of Crop Science
What this Specialist Model actually does
Decision: What will crop yield be for this field or planting, given current growing conditions, inputs, and history?
The final taxonomy records 6 supporting or external components, but it doesn't promote an Oumi-owned specialist model for this engine.
Who uses this Specialist Model
- VP Agronomy
- Head of Crop Science
- Farm Operations Director
Where the specialist model sits
The final taxonomy records 6 supporting or external components, but it doesn't promote an Oumi-owned specialist model for this engine.
Remote-sensing ingestion pipeline processing satellite/drone imagery into vegetation-index features
Supporting component
How Oumi connects to your stack →
Growth-stage-aware forecast updating service
Supporting component
Growth-stage-aware forecast updating service that improves precision as the season progresses
Early-warning layer flagging underperforming fields relative
Supporting component
Early-warning layer flagging underperforming fields relative to forecast for agronomist review
Read the Early Warning guide →
Explanation layer summarizing key yield drivers in plain language
Supporting component
Explanation layer summarizing key yield drivers in plain language for grower and buyer communication
Yield forecasting model combining historical yield, soil, and weather data, fine-tuned per crop and region
Supporting component
Post-harvest feedback capture comparing predicted versus actual yield
Supporting component
Post-harvest feedback capture comparing predicted versus actual yield to recalibrate the model each season
How to evaluate and improve it
Evaluation
- Yield prediction mean absolute percentage error against actual harvested yield
- Forecast accuracy by growth-stage checkpoint (early-season versus pre-harvest accuracy)
- Calibration of prediction confidence intervals against actual outcomes
Feedback signals
- Actual harvested yield compared against the predicted yield for the same field/block
- Forecast accuracy tracked across growth-stage checkpoints (does accuracy improve as harvest approaches, as expected)
- Agronomist override or manual adjustment of a forecast based on field-visit observations not captured in the data
- Weather events after a forecast that materially changed the actual outcome (used to evaluate forecast robustness, not as a model error)
- Buyer or contract volume shortfalls/surpluses traced back to yield forecast error


