🪨 Rock vs Mine Prediction using SONAR Data
📌 Project Overview
This project is a Machine Learning classification system that predicts whether an underwater object is a Rock or a Mine using SONAR signal data.
The project simulates a real-world submarine detection scenario, where sonar signals bounce differently from rocks and metal mines. A machine learning model is trained to identify this difference and make accurate predictions.
🎯 Problem Statement
Underwater mine detection is critical for submarine and naval safety. The goal of this project is to build a system that can automatically classify underwater objects as:
Rock (R)
Mine (M)
based on sonar signal readings.
🔄 Machine Learning Workflow
The project follows a standard ML pipeline:
Data Collection
SONAR dataset with 208 instances
Each instance has 60 numerical features
Binary target labels: Rock (R) or Mine (M)
Data Pre-processing
Data loaded using Pandas
Dataset does not contain column headers
Data prepared for training without missing values
Model Selection
Logistic Regression
Suitable for binary classification problems
Train-Test Split
Training set: 90%
Testing set: 10%
Ensures evaluation on unseen data
Model Evaluation
Accuracy score used to evaluate performance
🛠️ Technologies & Libraries Used:
Python
NumPy – Numerical computations
Pandas – Data handling and analysis
Scikit-learn
Logistic Regression
Train-Test Split
Accuracy Score
📊 Results
Training Accuracy: ~83.4%
Testing Accuracy: ~76%
The model performs well and generalizes reasonably on unseen data.
🔍 Predictive System
A simple predictive system is implemented where:
Raw sonar input data is provided
The model outputs:
“The object is a Rock” or
“The object is a Mine”
This demonstrates real-world usability of the trained model.
📁 Project Structure Rock-vs-Mine-Prediction/ │ ├── MinePrediction.ipynb ├── SolarData.csv └── README.md
🚀 How to Run the Project
Clone the repository:
git clone https://github.com/ankushx01-dev/SolarRock-VS-MinePredition-Project
Open the Jupyter Notebook:
jupyter notebook rock_vs_mine.ipynb
Run all cells sequentially.
📚 Learning Outcomes
Understanding of real-world ML workflows
Hands-on experience with classification problems
Practical use of Logistic Regression
Model evaluation using accuracy metrics
👤 Author
Ankush Rana B.E. CSE (AI & ML)
🙏 Acknowledgement
This project was developed for learning purposes by following an educational tutorial by Siddhardhan. The implementation, experimentation, and documentation were done independently to understand machine learning concepts in depth.
⭐ Future Improvements
Try advanced models like Random Forest or XGBoost
Perform feature scaling
Apply hyperparameter tuning
Improve accuracy and robustness