GIS-MCDA Slurry Suitability
R² = 0.77 (best model: random-forest regression kriging)
ProblemFarmers spread livestock slurry as fertiliser, but standard maps only say 'yes' or 'no' to a zone, ignoring how risk actually varies across a field.
SolutionA continuous, regulator-compliant suitability surface built with regression kriging, tested head-to-head against deep learning.
ImpactR² = 0.77, beating both the old Boolean overlay and fancier deep-learning alternatives.
Process simulation
Naziru explains this step
Full conversation walkthrough
A complete, in-depth explanation of this project, turn by turn. Runs about 5 minutes if you let it play.
Ask a question about this work
Send Naziru a question directly, it's delivered straight to his inbox.
Commonly asked questions
Private, for the author only.
This section is password-protected.
Incorrect password.