A Comparative Analysis of Different Pre-Processing Pipelines for EEG-Based Preference Prediction in Neuromarketing

Neuromarketing is an emerging brain-computer interface (BCI) research field that aims to understand consumers' internal decision-making processes when choosing which products to buy. It provides valuable insights for marketers to improve their marketing strategies based on consumers' impressions. However, the current status of Electroencephalography (EEG)-based preference prediction and its classification accuracy is still below optimal. The performance of an EEG-based preference prediction system largely depends on the suitable selection of a proper EEG preprocessing pipeline since noisy EEG data is likely not to give better results. Most of the studies followed a traditional EEG pre-processing pipeline and also recently an advanced automated EEG pre-processing pipeline has been introduced in the literature. In this study, we have proposed an optimal EEG pre-processing pipeline and compared it with the other two methods mentioned above. We collected raw EEG data and pre-processed them using the three pre-processing pipelines. We then extracted many statistical and frequency domain features and employed several machine-learning classification models. After implementing our proposed pipeline to preprocess EEG data, we observed a significant improvement in classification accuracies. Specifically, our pipeline achieved 100% accuracy for Ensemble learning, and 93.3% accuracy for k-Nearest Neighbor (kNN) model. In comparison, for the same classifiers the traditional and advanced automated pipelines achieved 75%, 83.3%, and 78.3%, 87% accuracies respectively.