Electroencephalographic (EEG) Dataset of Stress-Induced Auditory Stimulation for Event-Related Oscillation Analysis and Machine Learning Classification

This dataset contains 45 electroencephalographic (EEG) recordings acquired during an experimental protocol designed to induce stress corresponding to 15 subjects. The comparison conditions are relaxation and white noise auditory stimulation. The dataset was generated as part of the undergraduate thesis: “Analysis of Electroencephalographic Signals Associated with Stress Using Event-Related Oscillations and Machine Learning Techniques”by Yessica Rodriguez Hernandez at the Benemérita Universidad Autónoma de Puebla. The primary objective of this dataset is to support the analysis of oscillatory EEG activity associated with stress-related cognitive and emotional responses. EEG signals were analyzed using Event-Related Oscillations (ERO) based on Morlet wavelet time-frequency representations. Spectral power features extracted from multiple frequency bands were subsequently used for machine learning classification. The dataset includes: Raw and/or preprocessed EEG recordings. Experimental condition labels (stress, relaxation, white noise). Time-frequency representations and extracted spectral features (if included). Electrode configuration based on the international 10–20 EEG system. Metadata associated with signal acquisition and preprocessing. File names correspond to each condition, for example, "estres1raw." The prefix "estres" denotes the stress condition; the ordinal number corresponds to the subject ID, and "raw" means unprocessed data. There are 15 records for each condition. Preprocessed files do not have the label "raw" in the name. The experimental protocol was designed to evaluate changes in brain activity under different auditory stimulation conditions. Signal preprocessing included noise reduction and artifact removal procedures to improve EEG quality before feature extraction and classification. Potential applications of this dataset include: Stress detection using EEG. Brain-computer interface (BCI) research. Affective computing. Time-frequency EEG analysis. Event-related oscillation studies. Machine learning and pattern recognition applied to neurophysiological signals. Educational and research purposes in neuroscience, biomedical engineering, applied mathematics, and artificial intelligence. Machine learning methods explored in the associated thesis include Support Vector Machines (SVM), Linear Discriminant Analysis (LDA), Naive Bayes, and K-Nearest Neighbors (KNN). If this dataset is used in academic work, please cite the associated thesis.