Klasifikasi berbasis Regresi Menggunakan Hybrid Quantum Regression untuk Memprediksi Kualitas Teh Hijau

Ramadhani, Wendy Gata (2026) Klasifikasi berbasis Regresi Menggunakan Hybrid Quantum Regression untuk Memprediksi Kualitas Teh Hijau. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025221162-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Penilaian mutu teh hijau secara konvensional masih bergantung pada evaluasi sensori oleh panelis, sehingga hasilnya dapat dipengaruhi oleh subjektivitas, pengalaman penilai, dan variasi antar pengujian. Untuk mendukung proses penilaian yang lebih objektif, penelitian ini menerapkan klasifikasi berbasis regresi menggunakan data electronic nose (E-Nose) yang merekam profil aroma teh hijau melalui respons sensor. Model yang digunakan adalah Hybrid Quantum Regression (HQR), di mana nilai Aroma terlebih dahulu diprediksi sebagai keluaran kontinu, kemudian hasil prediksi tersebut dipetakan menjadi kelas mutu “Baik” atau “Cacat Mutu” menggunakan aturan threshold. Arsitektur model menggabungkan amplitude encoding untuk menyandikan fitur sensor ke dalam representasi kuantum dan data re-uploading untuk meningkatkan kemampuan model dalam mempelajari pola data. Hasil evaluasi menunjukkan bahwa model HQR Re-uploading memperoleh performa regresi yang baik dengan MAE sebesar 0,0121, RMSE sebesar 0,0281, dan R² sebesar 0,9889 pada data uji. Pada tugas klasifikasi berbasis regresi, model memperoleh accuracy sebesar 0,945, recall sebesar 1,000, dan F1-score sebesar 0,956. Hasil tersebut menunjukkan bahwa pendekatan HQR mampu memprediksi nilai Aroma secara efektif dan mendukung penentuan kualitas teh hijau secara lebih terukur, konsisten, dan berbasis data.
=====================================================================================================================================
Conventional green tea quality assessment still relies on sensory evaluation by panelists, which may be affected by subjectivity, assessor experience, and variation across evaluations. To support a more objective assessment process, this study applies regression-based classification using electronic nose (E-Nose) data that captures the aroma profile of green tea through sensor responses. The proposed model uses Hybrid Quantum Regression (HQR), in which the Aroma score is first predicted as a continuous output and then mapped into “Good” or “Defective Quality” classes using a threshold rule. The model architecture combines amplitude encoding to encode sensor features into a quantum representation and data re-uploading to improve the model’s ability to learn data patterns. The evaluation results show that the HQR Re-uploading model achieved strong regression performance, with a low prediction error and a high coefficient of determination, represented by an MAE of 0.0121, an RMSE of 0.0281, and an R² of 0.9889 on the test data. In the regression-based classification task, the model achieved an accuracy of 0.945, a recall of 1.000, and an F1-score of 0.956. These results indicate that the proposed HQR approach can predict Aroma scores effectively and support green tea quality assessment in a more measurable, consistent, and data-driven manner.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Data Re-uploading, Electronic Nose, Hybrid Quantum Regressor, Standar Kualitas Teh Hijau, Quantum Machine Learning, Data Re-uploading, Electronic Nose, Green Tea Quality, Hybrid Quantum Regressor, Quantum Machine Learning
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Wendy Gata Ramadhani
Date Deposited:	27 Jul 2026 06:26
Last Modified:	27 Jul 2026 12:32
URI:	http://repository.its.ac.id/id/eprint/138029

Actions (login required)

View Item