A new open-source implementation called the Turba fertilizer machine learning stack has been developed to enable reproducible site-specific fertilizer recommendations in Morocco. This three-layer system, comprising turba-client, turba-data, and turba-models, provides programmatic access to site profiles, crop-specific targets, and machine learning models for nutrient recommendations. The stack facilitates versioned outputs, independent loading and benchmarking of trained approximations, and links upstream retrieval with reproducible cross-model comparisons. The initial dataset includes 44,096 ESA WorldCereal locations and expands to over 132,000 crop-location recommendation requests, evaluating nine regression families to package the five best-performing crop-specific models. AI
IMPACT Provides a reproducible framework for agricultural nutrient recommendations, potentially improving crop yields and resource management.
RANK_REASON The item is a technical report detailing an open-source implementation for a specific application. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Daily Papers →
- ESA WorldCereal
- K_2O
- Morocco
- neutron
- P_2O_5
- turba-client
- turba-data
- Turba Fertilizer Machine Learning Stack
- turba-models
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →