Abstract—This paper proposed a method using principle component analysis based on graph entropy (PCA-GE) and J48 decision tree on electroencephalogram (EEG) signals to predict whether a person is alcoholic or not. Analysis is performed in two stages: feature extraction and classification. The principle component analysis (PCA) chooses the optimal subset of channels based on graph entropy technique and the selected subset is classified by the J48 decision tree in Weka. K-nearest neighbor (KNN) and support vector machine (SVM) in R package are also used for comparison. Experimental results show that the proposed PCA-GE method is successful in selecting a subset of channels, which contributes to the high accuracy and efficiency in the classification of alcoholics and non-alcoholics.
 

Index Terms—EEG, graph entropy, Horizontal Visibility Graph (HVG), Support Vector Machine (SVM), Principle Component Analysis (PCA), J48 decision tree

Cite: Shuaifang Wang, Yan Li, Pen Wen, and Guohun Zhu, "Analyzing EEG Signals Using Graph Entropy based Principle Component Analysis and J48 Decision Tree," International Journal of Signal Processing Systems, Vol. 4, No. 1, pp. 67-72, February 2016. doi: 10.12720/ijsps.4.1.67-72