The problem
Yield estimates come after harvest, too late for pricing and logistics.
What to build
Develop a model that estimates district yields during the season from satellite and weather inputs.
Success looks like
Mid-season error under an agreed percentage in back-tests over several seasons.
Suggested technologies
- Remote sensing features
- Gradient boosting
- Weather data
- Back-testing
- Uncertainty estimation
An original, independent statement. It is not an official Smart India Hackathon problem statement, and Arventiq Labs is not affiliated with or endorsed by that programme. The scenario is illustrative and does not describe any specific organisation. Free to use with credit.