Table 4 Comparative accuracy analysis for machine learning classifiers in Neuromarketing
From: Technological advancements and opportunities in Neuromarketing: a systematic review
Classifiers | Neuromarketing studies | Average accuracy |
|---|---|---|
Support Vector Machine (SVM) | Like/dislike classification for esthetic preference recognition among 3D objects (Chew et al.) [17] | 68% |
Attention bias identification between targeted and non-targeted stimuli using NeoCube-based SNN architecture (Doborjeh et al.) [64] | 48.5% | |
Like/dislike classification among e-commerce product (Yadava et al.) [18] | 62.85% | |
Emotional valence recognition between excitement and boredom using EEG device and combining SVM, KNN, SVR, LR (Ogino and Mitsukura) [68] | 72.4% | |
Purchase decision prediction from fMRI data using recursive cluster elimination-based support vector machine (RCE-SVM) (Wang et al.) [30] | 55.70% | |
Facial emotion recognition using GSR sensor biometric data (Goyal and Singh) [54] | 81.65% | |
Seven-emotion recognition using EEG signal (Bhardwaj et al.). Happiness and sadness classification accuracy reported here, respectively | 87.5%, 92.5% | |
Color classification using EEG signal (Rakshit et al.) | 78.81% | |
K-Nearest Neighbor (KNN) | Like/dislike classification for esthetic preference recognition among 3D objects (Chew et al.) [17] | 64% |
Hidden Markov model (HMM) | Like/dislike classification among e-commerce product (Yadava et al.) [18]. Classification accuracy reported for male and female subject, respectively | 70.33%, 63.56% |
Linear discriminant analysis (LDA) | Seven-emotion recognition using EEG signal (Bhardwaj et al.) [58]. Happiness and sadness classification accuracy reported here, respectively | 82.5, 87.5% |
Like-/dislike classification using car stimuli and ERP signal (Wreissenger et al.) | 61% | |
Naïve Bayes | Purchase decision prediction using Neural Impulse Actuator (NIA) device (Taqwa et al.) [73] | 48.5% |
Artificial Neural Network | Consumer gender prediction using facial action coding (Gurbuj and Toga) [28] | 83.8% |
TV advertisement liking recognition using EEG signal (Soria Morillo et al.) [43] | 80% | |
TV advertisement liking recognition using EEG (Soria Morillo et al.) [40] | 80% | |
Like/dislike classification among e-commerce products (Yadava et al.) [18] | 60% |