Sonar-Based Rock vs Mine Detection
A notebook experiment on binary classification of sonar signal measurements using logistic regression.
Problem
Can a supervised classifier separate the two labeled classes in a small, high-dimensional sonar dataset?
Logistic regression is used as the modeling approach in this experiment. The notebook is the primary evidence for preprocessing, training, and evaluation. The result should be understood as a dataset experiment, not as proof of performance in an operational detection system.
Interactive Notebook
Below is the complete Jupyter notebook for this project. You can expand/collapse code cells and view all outputs.
Loading notebook...
Product value and limitations
The notebook provides a compact way to inspect the full path from numeric sensor measurements to a binary prediction. The page does not claim field performance, sensor robustness, or deployment readiness. Those questions require independent test data, operating-condition coverage, and domain validation.