Abdalllah/crop-irrigation-mlflow

🌾 Intelligent Crop Irrigation Advisor - MLflow Dashboard

  

Live ML Experiment Tracking & Model Registry for Smart Agriculture 🚜

📊 Dashboard Overview

This MLflow server provides real-time tracking for 3 production-ready ML models that power an intelligent crop irrigation system:

🎯 Models Tracked

1️⃣ Crop Recommendation Model

Algorithm: Random Forest Classifier
Accuracy:  99.32%
Precision: 99.37%
Purpose:   Predict optimal crop for soil conditions
Features:  NPK levels, pH, Temperature, Humidity, Rainfall
Classes:   22 different crops

2️⃣ Irrigation Optimization Model

Algorithm: CatBoost Regressor  
R² Score:  0.974
MAE:       0.683 mm
RMSE:      2.603 mm
Purpose:   Calculate precise water amount needed
Features:  30+ engineered features (soil, weather, crop data)

3️⃣ Smart Irrigation Classifier

Algorithm: CatBoost + Optuna Hyperparameter Tuning
Accuracy:  83.61%
F1 Score:  83.57%
Purpose:   Binary classification - irrigate or not
Trials:    10 Optuna optimization runs tracked

🚀 How to Use This Dashboard

Navigate Through:

• —📁 Experiments → View all training runs with metrics & parameters
• —📊 Compare → Side-by-side comparison of multiple runs
• —🏆 Models → Model Registry with versioning
• —CropRecommendationModel (v2)
• —IrrigationOptimizationModel (v1)
• —SmartIrrigationClassifierModel (v1)

Quick Start:

1. 1.View Experiments: Click on any experiment name to see runs
2. 2.Compare Runs: Select multiple runs → Click "Compare"
3. 3.Download Models: Go to "Models" tab → Download .pkl files
4. 4.Visualize Metrics: Charts auto-generated for all logged metrics

� Key Features

✅ Real-time Tracking - Live experiment monitoring ✅ Parameter Comparison - Compare hyperparameters across runs ✅ Metric Visualization - Interactive charts for accuracy, loss, R² ✅ Model Registry - Versioned model storage with lineage ✅ Artifact Storage - Models, plots, and reports ✅ Optuna Integration - Hyperparameter optimization trials

📈 Metrics Dashboard

Tracked Metrics:

• —Classification: Accuracy, Precision, Recall, F1 Score
• —Regression: R² Score, MAE, RMSE, Adjusted R²
• —Feature Importance: Top features for each model
• —Training Time: Duration per run

Logged Parameters:

• —Model hyperparameters (learning_rate, depth, iterations)
• —Data split ratios (testsamples, trainsamples)
• —Feature engineering settings
• —Random seeds for reproducibility

🔗 Related Links

• —📚 GitHub Repository: Intelligent-Crop-Irrigation-Advisor
• —📖 Documentation: Full project docs in repo
• —🌱 Dataset: Agricultural data with soil & weather features
• —🔧 MLflow Guide: Setup instructions in MLFLOW_SETUP.md

� Tech Stack

📊 Model Performance Summary

🌍 Use Cases

• —🌾 Precision Agriculture: Optimize crop selection & water usage
• —💧 Water Conservation: Reduce water waste by 20-30%
• —📊 Farm Management: Data-driven irrigation decisions
• —🔬 Agricultural Research: Analyze model performance over time

🛠️ Local Development

Want to run this locally? Clone the repo:

git clone https://github.com/Iabdallllah/Intelligent-Crop-Irrigation-Advisor.git
cd Intelligent-Crop-Irrigation-Advisor
pip install -r requirements.txt
python mlflow_tools/launch_mlflow.py

Open: http://localhost:5000

📝 License

MIT License - See LICENSE for details

👨‍💻 Author

Abdallah - Agricultural AI Solutions 🔗 GitHub Profile

🙏 Acknowledgments

Built with passion for sustainable agriculture and smart farming 🌱

Using AI to grow more with less 💚

⚡ This Space is powered by Hugging Face Spaces + Docker + MLflow