Driver fatigue is a major issue since many people have been aware about safety degree of driving. In this regard, this paper proposes methods and application to determine the driving fatigue state for every 3 minutes. The collected EEG data come from 30 participants that were taken their EEG data using Emotiv EPOC+ with the duration of 33 or 60 minutes during driving simulation and their answers about the driving fatigue states for every 3 minutes. The participant and channel outliers were determined based on the correlation coefficient channels results with 3 highest correlation coefficient results ≤ 0,45 and the frequency of channels shown in the 3 highest correlation with counts ≥ 7. The data that have been determined their participant outliers will be grouped into class 1 (fit/alert) or class 2 (fatigue/sleepy). The preprocessing and classification will use the grouped data with the selected channels. The proposed method gives accuracy results using the KNN classification method with the maximum mean accuracy 96%, minimum accuracy 90%, and maximum accuracy 100%; and using the SVM classification method with a maximum mean accuracy 81%, minimum accuracy 60%, and maximum accuracy 90%.

Original languageEnglish
Pages (from-to)347-359
Number of pages13
JournalJournal of Theoretical and Applied Information Technology
Issue number3
Publication statusPublished - 30 Apr 2016


  • Driver fatigue prediction
  • Electroencephalogram (EEG)
  • Find significant channels
  • Find the best features
  • K-nearest neighbors (KNN)


Dive into the research topics of 'Classification of driver fatigue state based on EEG using Emotiv EPOC+'. Together they form a unique fingerprint.

Cite this