ZeniTsuIsSimP/npk-crop-recommendation

1---2title: NPK Crop Intelligence3emoji: 🌾4colorFrom: green5colorTo: yellow6sdk: streamlit7sdk_version: "1.28.0"8app_file: app/npk_crop_recommendation_app.py9pinned: false10license: mit11---12 13# 🌾 NPK Crop Intelligence — MLOps Edition14 15[![CI Pipeline](https://github.com/ZeniTsuisSimp/NPK-LIVE/actions/workflows/ci.yml/badge.svg)](https://github.com/ZeniTsuisSimp/NPK-LIVE/actions/workflows/ci.yml)16[![Hugging Face Spaces](https://img.shields.io/badge/🤗%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/ZeniTsuisSimp/npk-crop-recommendation)17![Python 3.11](https://img.shields.io/badge/python-3.11-blue.svg)18![MLflow](https://img.shields.io/badge/MLflow-tracked-orange.svg)19![DVC](https://img.shields.io/badge/DVC-versioned-purple.svg)20 21**Live Project:** https://huggingface.co/spaces/ZeniTsuisSimp/npk-crop-recommendation22 23 24Smart soil analysis and crop recommendation system powered by **Machine Learning**, built with **MLOps best practices** — DVC, MLflow, Docker, and CI/CD.25 26---27 28## 🏗️ Architecture29 30```mermaid31graph LR32    A[📊 Dataset] -->|DVC tracked| B[🔧 Preprocessing]33    B --> C[🤖 Training]34    C -->|MLflow logged| D[📈 Evaluation]35    D --> E[✅ Model .pkl]36    E --> F[🌐 Streamlit App]37    F -->|Docker| G[☁️ Deployment]38 39    subgraph "CI/CD"40        H[GitHub Push] --> I[Lint + Test]41        I --> J[Train Pipeline]42        J --> K[Docker Build]43        K --> L[Deploy]44    end45```46 47## 📂 Project Structure48 49```50NPK-LIVE/51├── .github/workflows/      # CI/CD pipelines52│   ├── ci.yml               # Lint, test, train on push/PR53│   └── cd.yml               # Docker build + deploy on main54├── src/                     # Modular ML pipeline55│   ├── data_preprocessing.py56│   ├── train.py             # MLflow-integrated training57│   └── evaluate.py          # Metrics generation58├── app/                     # Streamlit application59│   └── npk_crop_recommendation_app.py60├── tests/                   # Unit tests61│   └── test_pipeline.py62├── data/                    # Raw dataset63│   └── Crop_recommendation.csv64├── models/                  # Trained model artifacts65│   └── npk_crop_model.pkl66├── notebooks/               # Jupyter notebooks67├── reports/                 # Auto-generated metrics68├── params.yaml              # Hyperparameter config69├── dvc.yaml                 # DVC pipeline stages70├── Dockerfile               # Container config71├── docker-compose.yml       # Docker Compose72├── requirements.txt         # Production deps73└── requirements-dev.txt     # Dev/test deps74```75 76## 🚀 Quick Start77 78### 1. Clone & Install79```bash80git clone https://github.com/ZeniTsuisSimp/NPK-LIVE.git81cd NPK-LIVE82pip install -r requirements.txt83pip install -r requirements-dev.txt84pip install -e .85```86 87### 2. Run the ML Pipeline88```bash89# Step-by-step90python -m src.data_preprocessing91python -m src.train92python -m src.evaluate93 94# Or use DVC95dvc repro96```97 98### 3. Launch the App99```bash100streamlit run app/npk_crop_recommendation_app.py101```102 103### 4. Run with Docker104```bash105docker-compose up --build106# Visit http://localhost:8501107```108 109## 🔬 MLOps Stack110 111| Tool | Purpose |112|------|---------|113| **MLflow** | Experiment tracking (params, metrics, model artifacts) |114| **DVC** | Data versioning & reproducible pipelines |115| **Docker** | Containerized deployment |116| **GitHub Actions** | CI (lint + test) and CD (Docker + deploy) |117| **pytest** | Unit testing |118| **flake8** | Code linting |119 120### View MLflow Experiments121```bash122mlflow ui123# Visit http://localhost:5000124```125 126### Run DVC Pipeline127```bash128dvc repro           # Run full pipeline129dvc metrics show    # View latest metrics130dvc dag             # Visualize pipeline DAG131```132 133## 🧪 Testing134```bash135pytest tests/ -v136```137 138## 🔄 CI/CD Pipeline139 140### CI (on every push/PR)1411. **Lint** — `flake8` syntax & style checks1422. **Test** — `pytest` unit tests1433. **Train** — Full preprocessing → training → evaluation1444. **Verify** — Model and metrics artifacts exist145 146### GitHub Secrets Required147- `HF_USERNAME` — Hugging Face username148- `HF_TOKEN` — Hugging Face Write Token149 150## 📊 App Features151 152- 🌾 **Crop Prediction** — ML-based crop recommendation from soil NPK values153- 🧪 **NPK Additions** — Calculate nutrient amendments for target crops154- 🔄 **Crop Rotation Advisor** — Science-based next-crop suggestions155- 💚 **Soil Health Score** — Composite 0-100 health rating with radar chart156- 📅 **Seasonal Calendar** — Indian Kharif/Rabi/Zaid crop calendar157- 📈 **NPK History** — Track soil readings over time158 159## 📋 Model Details160 161| Parameter | Value |162|-----------|-------|163| Algorithm | Random Forest Classifier |164| Features | N, P, K (mg/kg) |165| Crops | Rice, Wheat, Corn, Barley, Soybean, Cotton, Sugarcane, Tomato, Potato, Onion |166| Config | `params.yaml` |167 168---169 170*Built with ❤️ using Streamlit, scikit-learn, MLflow, and DVC*