By requesting access to this dataset you agree to use the data responsibly and ethically in compliance with the CC BY 4 license - This dataset contains sensitive neurophysiological information and must be handled with care - You commit to using this data only for legitimate research educational or development purposes - You will respect the privacy and dignity of the participant who contributed this data - You will not attempt to identify or re-identify individuals from this data - You will give proper credit to Oliver Rösler and the UCI Machine Learning Repository in any work project or publication that uses this data - You will handle neurophysiological data responsibly and in accordance with ethical research standards

EEG Eye State Classification Dataset

Binary Classification of Eye States from EEG Signals

Overview

This dataset contains EEG measurements from 14 electrodes collected during different eye states (open/closed). It consists of 14,980 observations from a single individual, making it ideal for binary classification tasks in biosignal processing and brain-computer interface applications.

Key Features:

14 EEG channel measurements (international 10-20 system)
Binary target variable (eye open/closed)
Sampling rate: 128 Hz
Duration: ~117 seconds continuous recording
Device: Emotiv EEG Neuroheadset

Dataset Summary

The EEG Eye State dataset provides continuous electroencephalography recordings from a single participant during alternating eye states. Each of the 14,980 instances represents a snapshot of brain activity captured at 128 Hz across 14 electrodes positioned according to the international 10-20 system. The dataset is particularly valuable for researchers and practitioners working on biosignal classification, as it offers a clean, well-structured example of physiological time-series data with clear binary labels.

This dataset serves as an excellent benchmark for testing machine learning algorithms on real-world EEG data, with applications ranging from basic classification exercises to advanced brain-computer interface development.

Dataset Structure

Data Fields

Field	Description	Type
`AF3`, `F7`, `F3`, `FC5`, `T7`, `P7`, `O1`	EEG measurements from left hemisphere	Numeric
`O2`, `P8`, `T8`, `FC6`, `F4`, `F8`, `AF4`	EEG measurements from right hemisphere	Numeric
`eyeDetection`	Eye state (1 = closed, 0 = open)	Binary

Electrode Placement

The 14 EEG channels follow the international 10-20 system for electrode placement:

International 10-20 system for EEG electrode placement. The 14 channels in this dataset correspond to specific positions on the scalp.

Available Formats

ARFF Format: Original Weka format (EEG_Eye_State.arff)
CSV Format: Standard format for easy integration (EEG_Eye_State.csv)

Important Note: The data provided in both formats is raw and unprocessed, directly as captured from the Emotiv EEG Neuroheadset. The only transformation applied was the conversion from ARFF to CSV format for accessibility purposes. No preprocessing, filtering, normalization, or any other data manipulation has been performed. Users receive the original signal values as recorded during the data collection session.

Collection Methodology

Data Acquisition

Participant: Single individual
Device: Emotiv EEG Neuroheadset (consumer-grade wireless device)
Electrode System: 14 channels following the international 10-20 system
Sampling Rate: 128 Hz (128 measurements per second)
Recording Duration: Approximately 117 seconds of continuous data
Protocol: Participant alternated between opening and closing eyes during the session

Ground Truth Labeling

Eye states were recorded simultaneously with EEG measurements using a camera-based eye detection system, providing precise ground truth labels for each time point. This synchronization ensures high-quality annotations for supervised learning tasks.

Usage

Quick Start

import pandas as pd

# Load data
df = pd.read_csv('EEG_Eye_State.csv')

# Separate features and target
X = df.drop('eyeDetection', axis=1)
y = df['eyeDetection']

Common Applications

Binary classification of eye states
Time-series analysis and pattern recognition
Feature extraction from biosignals
Brain-computer interface research
Educational ML projects

Example Notebooks

🔗 AutoML Classification Example on Kaggle - Comprehensive notebook demonstrating various machine learning models and AutoML comparison for eye state classification.

Important Considerations

⚠️ Temporal Dependencies: Data is sequential. Use time-aware splitting to avoid data leakage.

⚠️ Single Subject: Collected from one individual. Generalization may require calibration.

⚠️ Class Balance: Check distribution before training. Use appropriate metrics (F1, ROC-AUC).

Citation

Original Publication:

Rösler, O. (2013). EEG Eye State. UCI Machine Learning Repository. https://doi.org/10.24432/C57G7J

BibTeX:

@misc{roesler2013eeg,
  author = {Rösler, Oliver},
  title = {EEG Eye State Dataset},
  year = {2013},
  publisher = {UCI Machine Learning Repository},
  url = {https://archive.ics.uci.edu/dataset/264/eeg+eye+state}
}

Acknowledgments

Original data collection: Oliver Rösler
Dataset hosting: UCI Machine Learning Repository
This dataset was originally sourced from the UCI Machine Learning Repository and is made available on Hugging Face for easier accessibility and integration with modern machine learning workflows.

License

CC BY 4.0 - Free to share and adapt with attribution

Downloads last month: 4

You need to agree to share your contact information to access this dataset