Klasifikasi Pola Atensi Pelanggan Terhadap Video Reklame Menggunakan Machine Learning Berbasis Sinyal Eeg: Studi Neuromarketing

Yahya, Rais (2024) Klasifikasi Pola Atensi Pelanggan Terhadap Video Reklame Menggunakan Machine Learning Berbasis Sinyal Eeg: Studi Neuromarketing. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022221008-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Perhatian adalah proses kognitif kompleks yang penting untuk fungsi sehari-hari. Penggunaan Elektroensefalogram (EEG) memungkinkan pengukuran perhatian karena resolusi temporalnya yang tinggi. Meskipun beberapa antarmuka otak-komputer (BCI) untuk pengenalan perhatian telah diusulkan, penelitian dengan jumlah partisipan substansial, paradigma yang tepat, dan analisis yang konsisten masih kurang. Penelitian ini meneliti dampak kualitas rangsangan visual terhadap sinyal EEG dan keterlibatan kognitif. Penelitian ini melakukan eksperimen untuk menyelidiki perhatian dan non-perhatian peserta selama pengumpulan data, baik ketika diberi rangsangan maupun tidak (baseline). Nilai Rata-rata Mutlak (MAV) dan Standar Deviasi (STD) digunakan untuk menilai perhatian sinyal EEG peserta terhadap variasi rangsangan visual.
Hasil menunjukkan penurunan aktivitas otak yang signifikan di area oksipital ketika individu memberikan perhatian tinggi dibandingkan perhatian rendah. MAV dan STD secara konsisten menurun selama keterlibatan visual aktif dibandingkan kondisi baseline tanpa rangsangan. MAV dan STD berkurang rata-rata 22% selama perhatian rendah dibandingkan baseline, sementara selama perhatian tinggi, kedua fitur menurun sebesar 53%. Hasil ini mungkin disebabkan oleh event-related desynchronization (ERD), yang menunjukkan penurunan amplitudo sinyal EEG dalam rentang frekuensi tertentu selama fokus tinggi.
Lebih lanjut, hasil klasifikasi menggunakan dua metode Naive Bayes mendukung temuan ini. Gaussian Naive Bayes menunjukkan akurasi pelatihan 81.5% dan akurasi uji 80%, sementara Kernel Naive Bayes menunjukkan akurasi pelatihan 87% dan akurasi uji 80%. Kedua metode efektif dalam membedakan kondisi atensi rendah dan tinggi. Gaussian Naive Bayes menunjukkan keseimbangan baik antara pelatihan dan uji, sedangkan Kernel Naive Bayes menunjukkan kemampuan lebih tinggi dalam mempelajari data pelatihan tanpa mengorbankan generalisasi. Hasil klasifikasi ini menegaskan bahwa perubahan signifikan dalam MAV dan STD dapat digunakan untuk mendeteksi kondisi atensi, menguatkan validitas analisis sinyal EEG dalam mengukur perubahan kognitif terkait atensi.
===================================================================================================================================
Attention is a complex cognitive process that is crucial for our daily functioning. Using Electroencephalogram (EEG) allows us to measure and assess attention because of its high temporal resolution. Although some brain-computer interfaces (BCI) for attention recognition have been proposed, there is a lack of research with a substantial number of participants, appropriate paradigms, and consistent analysis across individuals. This study examines the impact of visual stimulus quality on EEG signals and cognitive engagement. We conducted experiments to investigate participants' attention and non-attention states during data collection, both with and without visual stimuli (baseline). Mean Absolute Value (MAV) and Standard Deviation (STD) were used to assess the attention level in EEG signals concerning variations in visual stimuli. Our results show a significant decrease in brain activity in the occipital area when individuals paid high attention compared to low attention. Both MAV and STD consistently decreased during active visual engagement compared to the baseline condition without stimuli. MAV and STD decreased by an average of 22% during low attention compared to baseline, while during high attention, both features decreased by 53%. This may be due to a phenomenon known as event related desynchronization (ERD), which indicates a reduction in EEG signal amplitude within certain frequency ranges during high focus or attention. Furthermore, classification results using two Naive Bayes methods support these findings. Gaussian Naive Bayes showed a training accuracy of 81.5% and a test accuracy of 80%, while Kernel Naive Bayes showed a higher training accuracy of 87% with the same test accuracy of 80%. This shows that both methods are effective in distinguishing between low and high attention conditions. Gaussian Naive Bayes demonstrated a good balance between training and testing, while Kernel Naive Bayes showed a higher ability to learn from training data without sacrificing generalization ability. These classification results confirm that significant changes in MAV and STD can be used to detect attention conditions, strengthening the validity of EEG signal analysis in measuring cognitive changes related to attention.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Neuromarketing, EEG, Respons Otak, Atensi Konsumen Neuromarketing, EEG, Brain Response, Consumer Attention
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Yahya Rais
Date Deposited:	23 Jul 2024 08:59
Last Modified:	23 Jul 2024 08:59
URI:	http://repository.its.ac.id/id/eprint/108540

Actions (login required)

View Item