Article Per Year (5 Year)
p-Index From 2021 - 2026
0.92
P-Index
This Author published in this journals
All Journal Jurnal Sains Teknologi dan Sistem Informasi Jurnal Minfo Polgan (JMP) Jurnal Sains dan Teknologi
Kevin Bastian Sirait
Universitas Pelita Harapan
Aerospace Engineering Agriculture, Biological Sciences & Forestry Automotive Engineering Chemical Engineering, Chemistry & Bioengineering Chemistry Computer Science & IT Control & Systems Engineering Dentistry Electrical & Electronics Engineering Energy Engineering Health Professions Industrial & Manufacturing Engineering Library & Information Science Mathematics Mechanical Engineering Medicine & Pharmacology Nursing Public Health Social Sciences Veterinary

Published : 5 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 5 Documents
Search

 1 
Analisis Sentimen Komentar Toxic pada Video Musik YouTube Menggunakan Metode Naive Bayes Romindo Romindo; Kevin Bastian Sirait; Valentino Riffan; Chailine Garcia Wijaya
Jurnal Minfo Polgan Vol. 15 No. 1 (2026): Artikel Penelitian
Publisher : Politeknik Ganesha Medan

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.33395/jmp.v15i1.16190

Abstract

Platform berbagi video YouTube telah menjadi ruang interaksi digital yang masif, namun seiring pertumbuhannya, fenomena komentar toxic atau negatif semakin marak ditemukan pada konten-konten populer yang bersifat kontroversial. Penelitian ini bertujuan untuk menganalisis pola sentimen komentar toxic pada video musik "Hozier - Take Me to Church" serta mengukur performa algoritma Naive Bayes dalam melakukan klasifikasi sentimen secara otomatis. Data dikumpulkan menggunakan YouTube API dengan total 100.000 data komentar berbahasa Inggris, yang setelah melalui proses pembersihan menghasilkan 51.348 data valid. Proses text preprocessing mencakup lowercasing, penghapusan URL, tanda baca, angka, dan emotikon, diikuti tahap contraction, tokenization, penghapusan stopwords, serta lemmatization menggunakan library Natural Language Toolkit (NLTK). Data kemudian dilabeli ke dalam tiga kelas sentimen: positif, negatif, dan netral, sebelum diklasifikasi menggunakan algoritma Naive Bayes. Evaluasi performa model dilakukan dengan confusion matrix yang menghasilkan nilai akurasi sebesar 74%. Presisi untuk kelas negatif mencapai 96%, netral 97%, dan positif 64%. Nilai recall kelas negatif sebesar 57%, netral 49%, dan positif 99%. Sedangkan F1-score kelas negatif sebesar 71%, netral 65%, dan positif 78%. Dari total data yang diproses, ditemukan 8.475 komentar toxic (negatif), menunjukkan bahwa sebagian besar audiens merespons video tersebut secara positif. Hasil ini membuktikan bahwa algoritma Naive Bayes memiliki kemampuan yang memadai dalam klasifikasi sentimen komentar media sosial.
Klasifikasi Tingkat Kelunturan Warna Kain Menggunakan KNN, SVM, dan Random Forest Romindo Romindo; Triandes Sinaga; Kevin Bastian Sirait; Arosochi Yosua Daeli; Jepronel Saragih
INSOLOGI: Jurnal Sains dan Teknologi Vol. 5 No. 3 (2026): Juni 2026
Publisher : Yayasan Literasi Sains Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55123/insologi.v5i3.8721

Abstract

The laundry industry faces challenges in maintaining service quality, particularly regarding fabric color fading after washing. Assessments that are still performed manually tend to be subjective and inconsistent, so a more objective automated classification system is required. This study aims to apply and compare three algorithms, namely KNN, SVM, and Random Forest, to classify the level of fabric color fading based on digital images. The features used comprise color in the RGB and HSV spaces as well as shape in the form of area and shape ratio, all extracted automatically. A total of 300 images were divided into 250 training data and 50 testing data, then mapped into three categories, namely not faded, fairly faded, and faded. The testing results show that Random Forest delivers the best performance with an accuracy of 0.96, followed by SVM at 0.94 and KNN at 0.88. All models faced difficulties in recognizing the minority class due to data imbalance. This study proves that the machine learning approach, particularly Random Forest, is able to assess color fading levels more accurately and consistently than manual evaluation, while supporting quality control in the laundry industry.
Segmentasi Investor Cryptocurrency Menggunakan Metode K-Means: Studi terhadap Faktor-Faktor yang Mendorong Keputusan Investasi Arosochi Yosua Daeli; Vicky Darmana; Kevin Bastian Sirait; Triandes Sinaga
SATESI: Jurnal Sains Teknologi dan Sistem Informasi Vol. 5 No. 2 (2025): Oktober 2025
Publisher : Yayasan Pendidikan Penelitian Pengabdian ALGERO

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.54259/satesi.v5i2.7405

Abstract

The growth of cryptocurrency investors is very rapid with diverse characteristics and various investment driving factors. This study aims to analyze and form investor segmentation in cryptocurrency based on investment driving factors using the K-Means Clustering algorithm. A quantitative approach was applied through online questionnaires to 300 respondents who are cryptocurrency investors, with 289 valid data meeting the research criteria. The variables studied include four driving factors: Fear of Missing Out (FOMO), social media influence, high profit potential, and interest in the investment world. Data were processed through Min-Max normalization, Principal Component Analysis (PCA), and K-Means clustering using Orange Data Mining. The optimal number of clusters was determined using the Silhouette Score, while cluster validation used K-Nearest Neighbors (KNN). ANOVA and Games-Howell tests confirmed significant differences between clusters. The results identified four clusters: Cluster 1 (Emotional Investors, n=37), Cluster 2 (Ambitious Investors, n=156), Cluster 3 (Rational Investors, n=50), and Cluster 4 (Passive Investors, n=46). Cluster 3 is the most optimal in investment decision-making with a profit rate of 90% and zero loss (0%). These findings confirm that optimal investment decisions are driven by rational analysis and logical consideration without excessive emotional influence.
Analisis Kualitas Wine Menggunakan Machine Learning dengan Pendekatan SMOTE dan Seleksi Fitur Triandes Sinaga; Kevin Bastian Sirait; Jefri Junifer Pangaribuan; Okky Putra Barus; Romindo Romindo
INSOLOGI: Jurnal Sains dan Teknologi Vol. 4 No. 3 (2025): Juni 2025
Publisher : Yayasan Literasi Sains Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55123/insologi.v4i3.5436

Abstract

Conventional wine quality assessment remains reliant on subjective expert judgment, which introduces potential bias and inconsistency in quality control processes. This study aims to develop an objective and automated machine learning-based classification model to enhance the accuracy of wine quality prediction. To address the issue of class imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was applied, along with ANOVA F-test-based feature selection to optimize model performance. The White Wine Quality dataset from the UCI Machine Learning Repository (4,898 samples, 11 numerical features) was utilized to evaluate five classification algorithms: Naïve Bayes, Decision Tree, Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). Before SMOTE application, the Random Forest model achieved an accuracy of only 67.55%. After implementing SMOTE and parameter tuning, the Random Forest (Tuned) model demonstrated the best performance with 90.29% accuracy, 89.99% precision, 90.29% recall, and 89,97%.  % F1-score. Additionally, Decision Tree and KNN algorithms also exhibited notable improvements. SMOTE effectively balanced extreme minority class representations (quality levels 3 and 9). The most influential features in quality classification were alcohol content, density, and chlorides. These findings indicate that the proposed framework offers a reliable, objective, and scalable solution for automated wine quality control in industrial production environments.
Kuantifikasi Prioritas Pengembangan Fitur Aplikasi dengan Kerangka Outcome-Driven Innovation: Studi Kasus Aplikasi Pembelajaran Bahasa Asing Kevin Bastian Sirait; Nicholas Dickson; Romindo Romindo
INSOLOGI: Jurnal Sains dan Teknologi Vol. 4 No. 6 (2025): Desember 2025
Publisher : Yayasan Literasi Sains Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.55123/insologi.v4i6.6762

Abstract

This study aims to identify user satisfaction, importance, and development priorities for the English Language Speech Assistant (ELSA) application by integrating Outcome-Driven Innovation (ODI) and sentiment analysis methods. Using reviews collected from the Google Play Store via ScrapeStorm, two sentiment models were applied. Namely, Valance Aware Dictionary and sEntiment Reasoner (VADER) for overall sentiment polarity and Aspect-Based Sentiment Analysis (ABSA) for feature-spesific sentiment extraction. The result show that 86.23% of user reviews express positive sentiments, reflecting high satisfaction with ELSA’s functionality and learning experience. Correlation testing confirms strong validity between VADER and ABSA models (r = 0.98), and moderate alignment with user rating (r  0.55), confirming both construct and external validity. The ODI analysis further identifies Pronunciation as the highest-priority aspect (opportunity = 12.1), followed by Learning (opportunity = 2.6), highlighting key areas for innovation. Other aspects, such as Support and Subscription, show balanced performance with low opportunity values. The integration of ODI and sentiment analysis provide a data-driven framework for product improvement, enabling developers to prioritize enhancements based on user-perceived value. The findings contribute to informed strategic decision-making in the development of digital language-learning applications.
Co-Authors Arosochi Yosua Daeli Barus, Okky Putra Chailine Garcia Wijaya Jepronel Saragih Nicholas Dickson Pangaribuan, Jefri Junifer Romindo Romindo Triandes Sinaga Valentino Riffan Vicky Darmana