Implementasi Hybrid Collaborative Filtering untuk Rekomendasi Produk Penawaran pada Sistem Pengelolaan Dokumen CV. EKHABIMA

Susanto, Aloysius De Deo Darmo (2026) Implementasi Hybrid Collaborative Filtering untuk Rekomendasi Produk Penawaran pada Sistem Pengelolaan Dokumen CV. EKHABIMA. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025221174-Undergraduate_Thesis.pdf - Accepted Version Download (7MB)

Abstract

Penelitian ini bertujuan untuk mengembangkan dan mengevaluasi model rekomendasi Hybrid Collaborative Filtering yang mampu merekomendasikan produk secara relevan pada proses penyusunan dokumen penawaran di CV. Ekhabima, guna mengatasi terlewatnya peluang penawaran produk pelengkap. Model ini kemudian diintegrasikan ke dalam sistem informasi penawaran dan pengelolaan dokumen berbasis website menggunakan kerangka kerja Laravel, sebagai sarana implementasi yang memastikan moprodel dapat digunakan secara efektif dalam aktivitas operasional perusahaan sekaligus mengatasi inefisiensi pencatatan manual. Evaluasi performa model menunjukkan hasil kuantitatif yang sangat baik. Pada pengujian algoritma menggunakan metode Leave-One-Out (K=5), model Hybrid mencapai nilai Hit Rate sebesar 97,67% dan Precision sebesar 40,00%, melampaui threshold yang ditetapkan. Pendekatan Hybrid secara efektif mengatasi masalah bias popularitas pada matriks data yang memiliki tingkat sparsity 86,0%, dibuktikan dengan peningkatan metrik Coverage dari 3,30% menjadi 6,55% dibandingkan metode murni, sehingga memberikan peluang penawaran produk pelengkap pada produk berpermintaan rendah (long-tail). Ketahanan sistem terhadap kondisi edge cases mencapai 100% tanpa menghasilkan rekomendasi kosong. Pengujian waktu respons API layanan rekomendasi tercatat sangat optimal pada rentang 9 hingga 747 milidetik. Dari sisi sistem informasi sebagai sarana implementasi, Blackbox testing menunjukkan tingkat keberhasilan 100% pada 13 skenario use case, dan pengukuran System Usability Scale (SUS) memperoleh skor 77,5 yang masuk dalam kategori sangat baik. Implementasi ini juga terbukti memberikan penghematan waktu penyusunan draf penawaran hingga lebih dari 50% dibandingkan metode manual, sehingga meningkatkan efisiensi operasional.
======================================================================================================================================
This research aims to develop and evaluate a Hybrid Collaborative Filtering recommendation model capable of providing relevant product recommendations during the quotation drafting process at CV. Ekhabima, in order to address missed opportunities for proposing complementary products. The model is then integrated into a web-based quotation and document management information system built using the Laravel framework, serving as an implementation medium that ensures the model can be used effectively in the company's operational activities while also addressing the inefficiencies of manual recording. System performance evaluation demonstrates excellent quantitative results. In the algorithm testing using the Leave-One-Out method (K=5), the Hybrid model achieved a Hit Rate of 97.67% and a Precision of 40.00%, exceeding the set thresholds. The Hybrid approach effectively overcomes the popularity bias problem in a data matrix with an 86.0% sparsity rate, evidenced by an increase in the Coverage metric from 3.30% to 6.55% compared to the pure collaborative method, thereby creating opportunities to recommend complementary products among low￾demand (long-tail) products. System robustness against edge cases reached 100% without generating empty recommendations. The API response time testing for the recommendation service was highly optimal, ranging from 9 to 747 milliseconds. From the standpoint of the information system as an implementation medium, Blackbox testing showed a 100% success rate across 13 use case scenarios, and usability measurement using the System Usability Scale (SUS) obtained a score of 77.5, classifying it as Excellent. This implementation has also been proven to save over 50% of the time required to draft quotations compared to manual methods, significantly improving operational efficiency.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Collaborative Filtering, Recommendation System, Hybrid Filtering, Document Management, Information System, Collaborative Filtering, Sistem Rekomendasi, Hybrid Filtering, Pengelolaan Dokumen, Sistem Informasi.
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.6 Management information systems T Technology > T Technology (General) > T58.62 Decision support systems T Technology > T Technology (General) > T58.8 Productivity. Efficiency
Divisions:	Faculty of Information and Communication Technology > Informatics > 55201-(S1) Undergraduate Thesis
Depositing User:	Aloysius De Deo Darmo Susanto
Date Deposited:	26 Jul 2026 14:55
Last Modified:	26 Jul 2026 14:55
URI:	http://repository.its.ac.id/id/eprint/137612

Actions (login required)

View Item