Soil Analysis Web Application

This is a Django-based web application for analyzing soil data, predicting soil quality, and providing AI-driven crop suggestions. It features a user-friendly interface with a file upload system, interactive visualizations, and a chatbot powered by a reinforcement learning (RL) agent.

Features

• Soil Quality Prediction: Uses a Random Forest Classifier to predict soil grade (0-2) based on nutrient levels (N, P, K, pH, moisture) from uploaded CSV files.
• Interactive Visualizations: Displays nutrient levels as bar charts using Plotly.
• AI Suggestions: Provides initial rule-based suggestions and adaptive crop recommendations via a Q-learning RL agent.
• Chatbot: A popup chatbot on the right side, offering real-time assistance for soil and crop queries.
• File Upload: Allows users to upload CSV files for analysis, with results displayed on a dedicated page.

Prerequisites

• Python: 3.10.12
• Django: 5.0.4
• pandas:
• numpy:
• scikit-learn
• matplotlib:
• plotly:

Installation

1. Clone the Repository:

   git clone https://github.com/alexander784/Soil_Analysis.git
   cd soil_analysis
   

2. Set Up a Virtual Environment:

   python3 -m venv venv
   source venv/bin/activate
   

3. Install Dependencies:

   pip install -r requirements.txt
   

4. Apply Migrations:

   python3 manage.py makemigrations
   python3 manage.py migrate
   

5. Run dev server:

   python3 manage.py runserver
   
   

Usage

1. Upload soil Data:

• Visit the homepage (/) to access the upload page.
• Upload a CSV file with columns: N, P, K, pH, moisture.
• Example CSV format:

  N,P,K,pH,moisture
  75,45,120,6.5,50
  

2. View Results

• After uploading, you’ll be redirected to the results page showing:
  • Soil grade
  • A bar chart of nutrient levels.
  • AI suggestions (rule-based and RL-agent-driven).

3. Interact with the Chatbot

• A popup chatbot appears on the right side.
• Ask questions like "nitrogen" or "crop" to get responses based on the latest analysis.

4. Analyze Another Sample
   • Click "Analyze Another Sample" to return to the upload page.

Technical Details

Soil Prediction

• Uses RandomForestClassifier from scikit-learn with synthetic labels based on nutrient thresholds.
• Trained on-the-fly with each upload (no pre-trained model).

RL Agent

• Implements Q-learning in utils.py (SoilAgent class).
• States: Soil grade and nitrogen range.
• Actions: Crop recommendations (e.g., wheat, corn).

Future improvements

• Add CSV input validation and error messages.
• Implement real user feedback for the RL agent.
• Enhance chatbot with NLP (e.g., integrate a model like BERT).

Contributing

Feel free to fork this repository, submit issues, or create pull requests. Contributions are welcome!