Datasets:
license: ecl-2.0
task_categories:
- tabular-classification
language:
- en
tags:
- public
- tabular
- signal
- education
pretty_name: Sonar Rock Mine
size_categories:
- n<1K
The Sonar Rock Mine is a well-known benchmark dataset in the field of machine learning and pattern recognition, originally collected for the purpose of distinguishing between mines (M) and rocks (R) using sonar signals.
Dataset Overview:
- File:
raw_sonar.csv - Number of Features: 60 continuous numerical values per instance
- Target Label: One class label at the end of each row, either:
M→ Mine (target object: simulated sea mine)R→ Rock (non-mine object)
- Total Instances: The dataset contains multiple rows (exact count depends on full file), each representing a single sonar signal return.
Feature Description:
Each of the 60 features represents the energy or reflected signal strength recorded across different frequency bands at successive time intervals. These values are the result of bouncing sonar signals off underwater objects.
- The values are normalized and typically range between 0.0 and 1.0.
- They capture the acoustic signature of an object based on how it reflects sonar pulses.
- The variation in these reflections helps differentiate between the smooth, hard surface of a rock and the more complex, resonant structure of a mine.
Class Distribution:
The dataset generally has a relatively balanced distribution between the two classes:
- Mine (M): Signals reflected from simulated mines (often cylindrical, metal objects).
- Rock (R): Signals from natural rock-like objects or geological formations.
In real-world applications, this imbalance may vary, but the standard version used in ML is fairly balanced.
Objective:
The primary goal with this dataset is to build a binary classification model that can accurately classify whether a given sonar signal corresponds to a mine or a rock, which is critical for naval safety and underwater navigation.
Key Characteristics:
| Property | Description |
|---|---|
| Task Type | Binary Classification |
| Input Type | Continuous numerical features (time-series-like signal returns) |
| Number of Features | 60 |
| Number of Classes | 2 (M = Mine, R = Rock) |
| Typical Use Cases | Classification algorithms, feature selection, noise analysis, neural networks, SVM, decision trees |
| Challenges | High similarity between some signals, potential overfitting due to high dimensionality relative to sample size |
Summary:
This dataset provides a realistic example of sensory data classification, where subtle differences in signal patterns must be detected to make accurate decisions. It's widely used for teaching and evaluating classification techniques, especially in scenarios involving signal processing and underwater object detection.