Lecture Notes: Rock vs. Mine Prediction Using Sonar Data

Introduction

• Channel Focus: Artificial Intelligence and Machine Learning
• Content Schedule:
  • Monday & Wednesday: Basics of ML concepts/ topics
  • Friday: Detailed use case/ project in ML
• Current Topic: Rock vs. Mine Prediction using Sonar Data

Tools and Setup

• Programming Language: Python
• Platform: Google Collaboratory (Cloud-based system, no installation required)
  • Only need Google Chrome
  • Connects to Google Drive
  • Offers good storage (12 GB RAM, usable storage)

Use Case Explanation

• Problem: Predict whether an object underwater is a mine or a rock using sonar data
  • Example: Submarine navigation and identification of underwater mines
• Data: Sonar signals reflecting off objects (rock vs. metal)
  • Collected in a lab setting using a sonar device
  • Data format: CSV (Comma Separated Values)

Workflow

1. Data Collection
2. Pre-processing Data
   • Analyze and split data into training and test sets
3. Training Machine Learning Model
   • Logistic Regression (Binary classification)
   • Supervised Learning Algorithm
4. Testing and Validation
5. Prediction

Implementation Steps

Data Collection

• Source: Kaggle or UC Machine Learning repositories
• Dataset: sonar_data.csv (link provided in the video description)
• Features: 60
• Labels: 'R' for rock, 'M' for mine

Data Pre-processing

• Libraries Used: Import Dependencies import numpy as np import pandas as pd from sklearn.model_selection import train_test_split from sklearn.linear_model import LogisticRegression from sklearn.metrics import accuracy_score
• Loading Data into Pandas DataFrame sonar_data = pd.read_csv(path_to_csv_file, header=None)
• Description of Data
  • 208 rows, 61 columns
  • 1st to 60th columns: Features
  • 61st column: Labels ('R' or 'M')

Data Analysis

• Checking basic statistics: data.describe()
• Count of 'R' and 'M': data[60].value_counts()
• Grouping data based on 'R' and 'M': data.groupby(60).mean()

Splitting Data

• Separate Data and Labels X = data.drop(columns=60) Y = data[60]
• Split into training and test sets X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.1, stratify=Y, random_state=1)

Model Training

• Logistic Regression Model model = LogisticRegression() model.fit(X_train, Y_train)

Model Evaluation

• Training data accuracy training_predictions = model.predict(X_train) training_accuracy = accuracy_score(Y_train, training_predictions) print(f"Training Accuracy: {training_accuracy}")
• Test data accuracy test_predictions = model.predict(X_test) test_accuracy = accuracy_score(Y_test, test_predictions) print(f"Test Accuracy: {test_accuracy}")

Making Predictions

• Create a predictive system def predictive_system(data_example): ... # Convert data example to numpy array ... # Reshape data to match model input ... # Predict using logistic regression model return prediction
• Example: Check for rock and mine using sample data sample_input = [...] # Some example data prediction = predictive_system(sample_input) if prediction[0] == 'R': print("The object is a Rock") else: print("The object is a Mine")

Conclusion

• Check the description for datasets and Collab file
• Practice writing python scripts on your own
• Post queries in comments for clarification