Article Per Year (5 Year)
p-Index From 2021 - 2026
2.533
P-Index
This Author published in this journals
All Journal Bulletin of Electrical Engineering and Informatics Jurnal Edukasi dan Penelitian Informatika (JEPIN) Jurnal Nasional Pendidikan Teknik Informatika (JANAPATI) Journal of Robotics and Control (JRC) Jurnal Inovasi dan Teknologi Pembelajaran (JINOTEP) Kajian dan Riset Dalam Teknologi Pembelajaran Indonesian Journal of Computing, Engineering, and Design Jurnal Teknik Informatika (JUTIF) SinarFe7 Jurnal Ilmiah Ilmu Terapan Universitas Jambi Asian Journal Science and Engineering Evrimata: Journal of Mechanical Engineering
Pebrianti, Dwi
Unknown Affiliation
Aerospace Engineering Arts Humanities Automotive Engineering Chemical Engineering, Chemistry & Bioengineering Civil Engineering, Building, Construction & Architecture Computer Science & IT Control & Systems Engineering Education Electrical & Electronics Engineering Engineering Industrial & Manufacturing Engineering Materials Science & Nanotechnology Mechanical Engineering Social Sciences

Published : 16 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 16 Documents
Search

 1   2 
Using Content-Based Filtering and Apriori for Recommendation Systems in a Smart Shopping System Pebrianti, Dwi; Ahmad, Denis; Bayuaji, Luhur; Wijayanti, Linda; Mulyadi, Melisa
Indonesian Journal of Computing, Engineering, and Design (IJoCED) Vol. 6 No. 1 (2024): IJoCED
Publisher : Faculty of Engineering and Technology, Sampoerna University

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35806/ijoced.v6i1.393

Abstract

This research is motivated by the increasing significance of online shopping platforms and the challenges faced by users in locating products that align with their preferences and requirements, which can significantly influence the sales performance of online retailers. Consequently, the primary objective of this study is to design and implement a recommendation system capable of identifying suitable products and forecasting the purchase frequency for various product combinations, while also integrating this recommendation system with a smart shopping platform. To achieve this objective, the research employs machine learning techniques, specifically content-based filtering and the Apriori algorithm. Content-based filtering is utilized to analyze user preferences and behavioral patterns related to visited products, while the Apriori algorithm is employed to evaluate support and confidence values for item set combinations, thereby generating frequency values for future transactions involving product combinations. Additionally, a smart shopping system is developed and integrated, enhancing the shopping experience through smartphone applications and streamlining the payment process to facilitate seamless product purchases. The research methodology involves data collection pertaining to products and user preferences, followed by several testing involving a sample group of user respondents. The results demonstrate that the developed recommendation system effectively delivers relevant product recommendations based on user preferences, achieving a confidence value up to 98%. Furthermore, the smart shopping system proves capable of independently assisting users throughout the transaction process, thereby enhancing overall user experience and convenience.
The Effect of Pin Length and Compressive Force in Double Side Friction Stir Welding on Bending Strength of AA1100 Sukma Satriawan; Setiawan, Agus; Pebrianti, Dwi; Binti MD. Zain, Zainah
Evrimata: Journal of Mechanical Engineering Vol. 01 No. 01, 2024
Publisher : PT. ELSHAD TECHNOLOGY INDONESIA

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.70822/evrmata.vi.16

Abstract

Many new welding methods have emerged to improve connection results, including friction stir welding (FSW). FSW is a welding method that is widely used in welding aluminium alloys. FSW method on AA1100 aluminium material has not yet obtained the maximum bending strength so it is necessary to study the improvement of the quality of FSW joints using the welding method on both sides or double side friction stir welding (DFSW). This study aims to determine the effect of pin length and downward force on double side friction stir welding (DFSW) on the bending strength of AA1100 aluminium . The independent variables of this study are pin length (1.5 mm, 2 mm, 2.5 mm) and downward force (30 kg, 35 kg, 40 kg, 45 kg). The controlled variables are shoulder diameter of 25 mm, machine table translational speed of 10 mm/min, spindle rotation speed of 1750 rpm, base plate temperature of 250ºC, and AA1100 plate thickness of 3.6 mm with butt joint type welding connection model. The method used in this research is experimental using the factorial design of experiment (DOE) data analysis method. The results of this study indicate that pin length and downward force have a significant effect on the bending strength of DFSW welded joints on AA1100. The maximum bending strength value of the welded joint was 289.59 MPa at a pin length variation of 2 mm and a compressive force of 35 kg. The percentage of weld defects including tunnel and flash in welded joints with maximum bending strength is identified as the least and the micro test results also show the least FeAl3 particle grains.
Effect of Coconut Shell-Based Active Carbon Adsorbent on Motorcycle Exhaust Gas Emissions Putra Gitama, Nahindi; Hidayat, Najmul; Pebrianti, Dwi
Evrimata: Journal of Mechanical Engineering Vol. 01 No. 03, 2024
Publisher : PT. ELSHAD TECHNOLOGY INDONESIA

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.70822/evrmata.v1i03.57

Abstract

This study focused on the utilization of active carbon derived from coconut shells as an adsorbent to reduce exhaust gas emissions in motorcycles. The research aimed to compare the exhaust emissions before and after installing active carbon in the muffler and to analyze its effect on the levels of CO, HC, and CO2 at different engine speeds. A laboratory experiment was conducted with varying masses of active carbon, and emission data were collected and analyzed using two-way ANOVA. The results demonstrated that with the use of 200 grams of active carbon, the CO emission decreased by 12.06%, HC by 16.96%, and CO2 by 9.17%. These reductions are attributed to the strong adsorptive properties of active carbon, which facilitated the physical and chemical separation of harmful gases. The study concluded that active carbon significantly reduces exhaust emissions, providing a practical solution for improving air quality in motorcycles. The findings offer an effective method for emission control that could be applied under various operating conditions, making it suitable for widespread implementation in emission-reduction systems for small engines.
EXPLAINABLE MACHINE LEARNING FRAMEWORK FOR HOTEL CUSTOMER LOYALTY PREDICTION USING TRANSACTIONAL BEHAVIORAL DATA Pranoto, Gatot Tri; Pebrianti, Dwi; Religia, Yoga; Agusalim, Lestari
Jurnal Ilmiah Ilmu Terapan Universitas Jambi Vol. 10 No. 3 (2026): Volume 10, Nomor 3, June 2026
Publisher : LPPM Universitas Jambi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22437/jiituj.v10i3.56180

Abstract

Customer loyalty has become a critical factor for sustaining competitiveness in the hotel industry, particularly in increasingly digital and data-driven business environments. Although hotels continuously generate large volumes of transactional customer data, transforming this data into actionable insights for customer retention and marketing decision-making remains a significant challenge. This study proposes an Explainable Machine Learning Framework for hotel customer loyalty prediction using transactional behavioral data. The study used a publicly available hotel customer transaction dataset from the Mendeley Data repository, comprising 2,000 customer records. A supervised machine learning approach was employed using Logistic Regression, Decision Tree, Random Forest, XGBoost, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Naïve Bayes. Model performance was evaluated using Stratified K-Fold Cross-Validation and classification metrics, including Accuracy, Precision, Recall, F1-Score, and ROC-AUC. Experimental results demonstrated that all evaluated models achieved perfect classification performance, with Accuracy, Precision, Recall, F1-Score, and ROC-AUC values reaching 1.000. SHAP analysis revealed that frequency_of_bookings, days_since_last_booking, total_meal_charges, and total_revenue_generated were the primary drivers of customer loyalty prediction, while average_stay_duration showed minimal influence. The findings indicate that booking frequency and customer recency are the most influential behavioral factors affecting loyalty outcomes. From a managerial perspective, the proposed framework provides actionable insights for customer retention, customer segmentation, loyalty programs, and personalized marketing strategies. This study contributes to predictive customer analytics and Explainable Artificial Intelligence by integrating predictive accuracy with transparent interpretation of customer behavior in hotel loyalty management.
Sistem Rekomendasi Jalur Sertifikasi Berbasis Machine Learning untuk Lifelong Learning Jody, Jody; Aryo Riandhito, Febry; Ridwan, Mohamad; Pebrianti, Dwi
JEPIN (Jurnal Edukasi dan Penelitian Informatika) Vol. 12 No. 2 (2026): Volume 12 No 2
Publisher : Program Studi Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Dinamika pasar tenaga kerja saat ini menuntut peningkatan kompetensi secara berkelanjutan melalui ekosistem lifelong learning (pembelajaran sepanjang hayat). Lembaga Sertifikasi Profesi (LSP) berperan penting dalam hal ini, namun sering kesulitan memberikan rekomendasi sertifikasi yang tepat bagi pendaftar baru. Kendala utamanya meliputi ketiadaan riwayat sertifikasi (cold-start), keterbatasan pencocokan kata kunci pada sistem tradisional (lexical mismatch), serta kurangnya transparansi alasan rekomendasi dari sistem kecerdasan buatan (black-box). Penelitian ini mengusulkan sistem rekomendasi hibrida yang menggabungkan Machine Learning dan Generative AI. Untuk mengatasi masalah cold-start, algoritma K-Means digunakan untuk mengelompokkan profil pendaftar ke dalam 6 persona karier yang optimal berdasarkan data awal mereka. Selanjutnya, pencocokan profil dengan skema sertifikasi dilakukan menggunakan model Sentence-BERT (SBERT) yang mampu memahami konteks makna bahasa, bukan sekadar kata persis. Hasil pengujian menunjukkan pendekatan hibrida ini sangat unggul dengan tingkat akurasi (Recall@3) mencapai 82,67%. Terakhir, untuk mengatasi masalah black-box, sistem ini mengimplementasikan teknologi Retrieval-Augmented Generation (RAG). Teknologi ini menerjemahkan skor komputasi menjadi penjelasan naratif yang mudah dipahami, memberi tahu pengguna mengenai kesenjangan keterampilan (skill gap) yang perlu mereka pelajari. Secara keseluruhan, prototipe sistem berbasis web ini berhasil memberikan rekomendasi yang presisi sekaligus bertindak sebagai penasihat karier transparan yang mendukung pengembangan kompetensi tenaga kerja.
Analisis Segmentasi Konsumen E-Commerce menggunakan Model RFM melalui Metode K-Means, DBSCAN, dan Agglomerative Clustering Diana, Putri; Yusup, Rika; Pebrianti, Dwi
JEPIN (Jurnal Edukasi dan Penelitian Informatika) Vol. 12 No. 2 (2026): Volume 12 No 2
Publisher : Program Studi Informatika

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Data transaksi perdagangan elektronik sering tersedia dalam jumlah besar, tetapi belum selalu diterjemahkan menjadi segmentasi pelanggan yang dapat mendukung keputusan pemasaran. Penelitian ini bertujuan membandingkan K-Means, DBSCAN, dan klasterisasi aglomeratif berbasis model Recency, Frequency, dan Monetary (RFM) serta menentukan metode yang paling sesuai dengan karakteristik data. Set data terdiri atas 34.500 transaksi dari Kaggle yang merepresentasikan 7.903 pelanggan. Tahapan penelitian mencakup analisis eksploratif data, prapemrosesan, pembentukan atribut RFM, standardisasi Z-Score, klasterisasi, serta evaluasi menggunakan Silhouette Score, Davies-Bouldin Index (DBI), dan Calinski-Harabasz Index (CHI). K-Means dan DBSCAN memperoleh Silhouette Score yang sama, yaitu 0,348. DBSCAN menghasilkan DBI lebih rendah (0,601), sedangkan K-Means memperoleh CHI tertinggi (4472,039) dan membentuk distribusi segmen yang lebih seimbang serta mudah diinterpretasikan. Oleh karena itu, K-Means dipilih sebagai metode yang paling aplikatif pada set data ini dan menghasilkan tiga segmen, yaitu 4.336 Pelanggan Potensial, 1.901 Pelanggan Terbaik, dan 1.666 Pelanggan Berisiko Tidak Aktif. Kontribusi penelitian ini terletak pada perbandingan tiga paradigma klasterisasi dalam kerangka RFM yang sama dan penerjemahan hasil klaster menjadi strategi loyalitas, pengembangan pelanggan potensial, serta aktivasi kembali pelanggan yang berisiko tidak aktif.
Co-Authors Addie Irawan Ahmad, Denis Andalas Saputra, Meitro Hartanto Aryo Riandhito, Febry Azahar, Mohd Iskandar Putra binti MD. Zain, Zainah Dian Maharani, Dian Fachri, Moch Fadhil Abas, Mohammad Faizal, Elka Fresy Nugroho Gatot Tri Pranoto Hadi Susilo, Sugeng Hammad, Jehad A.H. Hammad, Jehad AH Herwan Sulaiman, Mohd Hidayat, Najmul Huda, Arif Nur I Gusti Putu Asto Buditjahjanto Jody, Jody Lestari Agusalim Lestari, Tri Mukti Luhur Bayuaji, Luhur Melisa Mulyadi Mohamad Ridwan Mohd Saad, Norhafidzah Muhammad Ridho Nahindi Putra Gitama Nurrahma ‘N, Alfina Nurrahma, Alfina Prakasa, Aji Bagas Pratama, Maulana Krisna Bayu Putri Diana Roro Inda Melani Rusdah Sharif Zakaria, Mohd Sukma Satriawan Tri Ika Jaya Kusumawati Wijayanti, Linda Yoga Religia Yulia Puspita Dewi Yusup, Rika Zidan, Muhammad