🌱 Crop Recommendation API

A FastAPI-based Machine Learning API that recommends the top 5 crops suitable for cultivation based on soil and weather parameters. The model is powered by a RandomForest Classifier trained on the final preprocessed dataset (final_train_balanced).

---

📌 Features

• 🚀 FastAPI backend with RESTful endpoints

• 🌾 RandomForest model trained on 7 key soil & climate features:

  • Nitrogen (N)
  • Phosphorus (P)
  • Potassium (K)
  • Temperature
  • Humidity
  • pH
  • Rainfall

• 🔮 Returns Top-5 ranked crop recommendations with confidence scores

• 📦 Dockerized for easy deployment

• 🔑 JSON API for integration with frontend / Node.js backend

---

📂 Project Structure

crop-recommendation-api/
│── rf.pkl                # Trained RandomForest model
│── main.py               # FastAPI app with endpoints
│── requirements.txt      # Dependencies
│── Dockerfile            # For containerization
│── README.md             # Project documentation

---

⚙️ Installation

1️⃣ Clone the Repository

git clone https://github.com/yourusername/crop-recommendation-api.git
cd crop-recommendation-api

2️⃣ Create Virtual Environment & Install Dependencies

python -m venv venv
source venv/bin/activate   # For Linux/Mac
venv\Scripts\activate      # For Windows

pip install -r requirements.txt

3️⃣ Run FastAPI Server

uvicorn main:app --reload

Server runs at 👉 http://127.0.0.1:8000

---

🛠️ API Endpoints

1. Root Endpoint

GET /

Response:

{"message": "Crop Recommendation API is running 🚀"}

---

2. Predict Top-5 Crops

POST /predict

📥 Request Body (JSON):

{
  "N": 90,
  "P": 42,
  "K": 43,
  "temperature": 25.5,
  "humidity": 80,
  "ph": 6.5,
  "rainfall": 200
}

📤 Response:

{
  "recommendations": [
    {"crop": "rice", "probability": 0.65},
    {"crop": "maize", "probability": 0.20},
    {"crop": "banana", "probability": 0.07},
    {"crop": "mango", "probability": 0.05},
    {"crop": "cotton", "probability": 0.03}
  ]
}

---

🐳 Run with Docker

1️⃣ Build Docker Image

docker build -t crop-api .

2️⃣ Run Container

docker run -d -p 8000:8000 crop-api

Now API is available at 👉 http://localhost:8000

---

📊 Model Training

• Model: RandomForestClassifier (n_estimators=200, random_state=42)
• Dataset: final_train_balanced (7 numeric features + label)
• Output: rf.pkl saved with feature names

---

👨‍💻 Tech Stack

• FastAPI – RESTful API framework
• Scikit-learn – Machine Learning (RandomForest)
• Joblib – Model persistence
• Uvicorn – ASGI server
• Docker – Containerization

---

🚀 Future Enhancements

• 🌍 Add state & district level recommendations
• 📈 Improve ranking with ensemble models
• 🌦️ Integrate real-time weather APIs

---