The Soil Anomaly Detection Engine is an AI-powered system designed to detect chemical anomalies in soil, specifically focusing on identifying "chemical spiking" (e.g., sudden dumping of urea or other fertilizers) versus natural "organic" nutrient variations.

It uses an Isolation Forest model to analyze not just the current nutrient levels ($N, P, K, EC$) but also the rate of change (Deltas) from previous readings. This allows the system to differentiate between a naturally high-nutrient soil (slow change) and an artificial spike (fast change).

• Real-Time Anomaly Detection: Instantly flags readings that deviate from organic patterns.
• Delta Analysis: Calculates the speed of change for Nitrogen, Phosphorus, Potassium, and Electrical Conductivity (EC).
• Organic Score: Assigns a credit-score-like rating (0-100) to soil health.
• REST API: Exposes endpoints for analysis via FastAPI.
• Synthetic Data Generation: Simulates realistic organic and chemical fraud scenarios for training.

soil-anomaly-detection-engine/
├── dataset/                # Generated CSV datasets (train/test)
├── models/                 # Saved machine learning models (.pkl)
├── notebooks/              # Jupyter notebooks for experimentation
├── src/                    # Source code
│   ├── main.py             # FastAPI application entry point
│   ├── process_real_data.py# Data generation script (Organic vs Chemical)
│   ├── train_model.py      # Model training script
│   └── evaluate_model.py   # Model evaluation and testing script
├── requirements.txt        # Python dependencies
└── Readme.md               # Project documentation

1. Clone the repository:

   git clone <repository-url>
   cd soil-anomaly-detection-engine

2. Create a virtual environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

Follow these steps to generate data, train the model, and run the system.

Generate synthetic datasets (Training & Testing) that simulate both organic soil patterns and chemical spikes. This script reads Crop_recommendation.csv (if available) or generates synthetic baselines.

python src/process_real_data.py

Output: Creates dataset/train_data_final.csv and dataset/test_data_final.csv.

Train the Isolation Forest model on the generated dataset.

python src/train_model.py

Output: Saves the trained model to models/isolation_forest_final.pkl.

Run the evaluation script to test accuracy, view the confusion matrix, and simulate specific fraud scenarios (e.g., "Sudden Urea Spike").

python src/evaluate_model.py

Output: Displays accuracy score, classification report, and scenario test results.

Start the FastAPI server to accept real-time soil data requests.

uvicorn src.main:app --reload

The API will be available at http://127.0.0.1:8000.

POST /analyze_soil

Analyzes a soil reading and determines if it is organic or an anomaly based on previous history (in-memory).

Request Body:

{
  "farm_id": "farm_101",
  "Nitrogen": 120.0,
  "Phosphorus": 45.0,
  "Potassium": 50.0,
  "EC": 1.2
}

Response:

{
  "farm_id": "farm_101",
  "status": "COMPLIANT",
  "organic_score": 90,
  "details": {
    "is_first_reading": false,
    "anomalies_found": false,
    "confidence_score": 0.15
  },
  "sensor_summary": {
    "Nitrogen": "120.0 (Change: 2.0)",
    "EC": "1.2 (Change: 0.05)"
  }
}

MIT License (or specify your license here)