Tech Stack & Data Pipeline
- Data Ingestion & Cleaning: 25,000+ soil & weather rows via CSV → Pandas.
- Feature Engineering: Nutrient ratios, seasonal humidity trends, pH-interaction terms.
- Modeling: RF selected (R²=0.92, MAE=0.15) after benchmarking Linear, GBM.
- Validation: 5-fold CV, grid-search, SHAP explainability.
- Deployment (MLOps): Docker + GitHub Actions CI/CD → Render.com.
Live App Preview
In modern agriculture, data reigns supreme. By integrating N, P, K soil metrics and climatic variables into an advanced regression pipeline, this system delivers actionable crop recommendations to maximize yield.
Architected end-to-end: Jupyter → PyCharm → Docker → CI/CD → Render, enabling real-time, scalable predictions across diverse agro-climates.
Architected end-to-end: Jupyter → PyCharm → Docker → CI/CD → Render, enabling real-time, scalable predictions across diverse agro-climates.