Article Per Year (5 Year)
p-Index From 2020 - 2025
0.444
P-Index
This Author published in this journals
All Journal Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control Jurnal Media Teknik Elektro dan Komputer
Sitompul, Boy Arnol
Unknown Affiliation
Computer Science & IT Control & Systems Engineering Electrical & Electronics Engineering Energy Engineering

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Improving the Major Recommendation Systems: Analysis of Hybrid Naïve Bayes-based Collaborative Filtering and Fuzzy Logic Amir Saleh; Sitompul, Boy Arnol; Wijaya Laia, Laksana Febri; Sinaga, Nicholas Ferdinan
Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control Vol. 8, No. 4, November 2023
Publisher : Universitas Muhammadiyah Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22219/kinetik.v8i4`.1797

Abstract

Major recommendation systems have been widely used to assist prospective students in choosing major that matches their interests and potential. In an effort to improve the performance of the recommendation system, this study proposed to use collaborative filtering techniques with naïve Bayes approach. In addition, this study improved the input parameters using fuzzy logic in determining the recommended majors. The methodology used started from collecting user data, including gender, academic history, interests, and other relevant attributes. The data were used to train the naïve Bayes technique by estimating the probability of feature conformity between users and students in the recommended majors. However, there were problems such as uncertainty and ambiguity in user preferences for input data. The fuzzy logic method aimed to improve the input parameters to more accurately reflect the user preferences. The results of improving the input parameters by using fuzzy logic were then used in the naïve Bayes technique to obtain recommendations for the direction that best suits the user’s preferences. The final stage of this study used evaluation metrics such as precision, recall, and f1-score to measure the performance of the recommendation system in providing accurate recommendations. The use of a hybrid of naïve Bayes and fuzzy logic algorithms obtains an accuracy value of 87.27%, a precision value of 87.33%, a recall value of 87.24%, and an f1-score value of 87.26%. These results are higher than the usual naïve Bayes model applied in major recommendation systems.
Analisis Sentimen Komentar Intagram Pemindahan Ibu Kota Negara Membandingkan Alogritma Support Vecthor Meachine dan Random Forest Mesanda, Zery; Sitompul, Boy Arnol
Jurnal Media Teknik Elektro dan Komputer Vol 2 No 1 (2025): Metrokom : Jurnal Media Teknik Elektro dan Komputer
Publisher : Yayasan Pendidikan Al-Yasiriyah Bersaudara

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.65371/metrokom.v2i1.59

Abstract

Social media sentiment analysis has become an important approach in understanding public opinion on strategic issues, including the discourse on the relocation of the national capital. This study aims to compare the performance of Support Vector Machine (SVM) and Random Forest (RF) algorithms in classifying the sentiment of public comments on Instagram. A total of 794 comment data were collected using web scraping techniques with Selenium and BeautifulSoup, then divided into 80% training data and 20% test data. The classification process was conducted after the text preprocessing stage, which included case folding, tokenizing, filtering, and stemming. The experimental results show that SVM achieved an accuracy of 75.0% with precision 0.7200, recall 0.7800, and F1-score 0.7488. Meanwhile, Random Forest performed better with an accuracy of 79.4%, precision of 0.7795, recall of 0.8200, and F1-score of 0.7992. Evaluation based on sentiment class shows that SVM can only achieve a correct rate of 75.0% in the positive class and 75.1% in the negative class, while Random Forest excels with 79.4% in the positive class and 79.3% in the negative class. These findings confirm that Random Forest is more optimal and consistent than SVM in sentiment analysis based on social media comments. This study recommends the use of ensemble learning algorithms such as Random Forest in similar studies, as well as further development with larger datasets and deep learning approaches to improve model accuracy and generalization.
Co-Authors Amir Saleh Mesanda, Zery Sinaga, Nicholas Ferdinan Wijaya Laia, Laksana Febri