ABSTRACT

In this paper the performance of oversampling methods such as SMOTE (Synthetic Minority Over-sampling Technique) and PCA (Principal Component Analysis) which are used for preprocessing are applied for the Brain computer interface dataset. The pre-processed data is used for classification by SMO and Naïve Bayes. In the EEG recordings, the transient events are detected while predicting the conditions of Central Nervous System and are classified as epileptic spikes, muscle activity, eye blinking activity and sharp alpha activity. The Preprocessing technique SMOTE is an over-sampling method which combines the informed over-sampling of minority class with random under-sampling of the majority class. Principal Component Analysis (PCA) is an exploratory data analysis technique. It involves a mathematical procedure which transforms a number of possibly correlated variables into smaller number of uncorrelated variables called Principal Components. It is mostly used as a tool in data analysis and for making predictive models. Based on the experimental results derived through SMOTE and PCA when they are applied to SMO and Naïve Bayes, it is concluded that PCA can be a better option since its performance improvement is better than that of SMOTE.