cover
Contact Name
Mesran
Contact Email
mesran.skom.mkom@gmail.com
Phone
-
Journal Mail Official
jurnal.bits@gmail.com
Editorial Address
-
Location
Kota medan,
Sumatera utara
INDONESIA
Building of Informatics, Technology and Science
ISSN : 26848910     EISSN : 26853310     DOI : -
Core Subject : Science,
Building of Informatics, Technology and Science (BITS) is an open access media in publishing scientific articles that contain the results of research in information technology and computers. Paper that enters this journal will be checked for plagiarism and peer-rewiew first to maintain its quality. This journal is managed by Forum Kerjasama Pendidikan Tinggi (FKPT) published 2 times a year in Juni and Desember. The existence of this journal is expected to develop research and make a real contribution in improving research resources in the field of information technology and computers.
Arjuna Subject : -
Articles 926 Documents
Perbandingan Algoritma SVM, Random Forest, KNN untuk Analisis Sentimen Terhadap Overclaim Skincare pada Media Sosial X Rahmawati, Ira Tri; Alita, Debby
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6782

Abstract

The cosmetic industry in Indonesia, especially skincare products, is growing rapidly along with changes in people's lifestyles and technological advances. One of the main issues that arise is overclaiming, which can harm consumers and damage the company's reputation. This study aims to compare the performance of three algorithms in sentiment analysis of skincare overclaims on X social media. The evaluated algorithms include Support Vector Machine (SVM), Random Forest, and K-Nearest Neighbors (KNN). The research dataset consists of 7,774 tweets collected between October 1 and November 30, 2024, with 5,559 tweets after the preprocessing stage, consisting of 4,281 negative sentiment tweets and 1,275 positive sentiment tweets. Data imbalance was addressed using the Synthetic Minority Over-sampling Technique (SMOTE), with 80% data split for training and 20% for testing. The results showed that before the application of SMOTE, the Random Forest algorithm had the highest accuracy of 95%, followed by Support Vector Machine at 91% and K-Nearest Neighbors at 80%. After the application of SMOTE, the accuracy increased significantly, with Random Forest reaching 98%, Support Vector Machine 97%, and K-Nearest Neighbors 84%. Random Forest proved to be the best algorithm, with the highest performance before and after SMOTE implementation, and was effective in handling both sentiment classes. This research provides insights for the skincare industry and regulators to detect and address product over-claiming issues through machine learning-based approaches.
Perbandingan Algoritma Random Forest, KNN, SVM Untuk Analisis Sentimen Pengalaman Belanja Thrift Di X Raihandika, M Rafi; Suryono, Ryan Randy
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6797

Abstract

The thrifting phenomenon is gaining traction, especially among millennials and Generation Z. Along with the increasing interest in thrifting, X social media has emerged as one of the main platforms for people to share experiences and opinions related to thrift shopping. This research aims to analyze people's sentiments about thrift shopping experiences by comparing the performance of Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN) algorithms. The dataset used in this study was obtained from Twitter as many as 6,390 tweets collected through crawling techniques with a time span of August 2, 2024 to September 4, 2024. The dataset is then processed to produce clean data. After the cleaning process, the data is divided 80:20 for training and testing. In testing the three algorithms, an accuracy level is obtained that shows how well the model makes predictions. This accuracy measures the extent to which the model successfully predicts the sentiment of the thrifting shopping experience based on the Twitter dataset. The results show that the Random Forest algorithm has the highest accuracy with 95%, precision 97%, recall 78%, and f1-score 85%. SVM achieved 93% accuracy, 93% precision, 72% recall, and 78% f1-score. KNN obtained 89% accuracy, 72% precision, 59% recall, and 61% f1-score. From the results obtained, the Random Forest algorithm shows the best accuracy for sentiment analysis of thrifting experiences on Twitter Indonesia. Its advantage lies in its stable ensemble learning approach, where multiple decision trees are combined to produce more accurate predictions. This ability makes Random Forest effective in handling varied and complex Twitter text data, making it the most reliable algorithm in this context.
Analisis Klasifikasi Sentimen Prediksi Rating Aplikasi Apple’s AppStore Dengan Menggunakan Metode Algoritma Random Forest Harahap, Armyka Pratama; Karim, Abdul; Rohani, Rohani
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6812

Abstract

Users of Apple's App Store applications are increasingly widespread among smartphone users. However, user responses to these apps vary widely. In addition, continuous developments in adding features and editing capabilities have led to the increasing complexity of using these applications. This research aims to analyze the sentiment of application users on Apple's App Store through reviews on the Google Play Store using the Random Forest method. This method was chosen to efficiently identify and group user responses into positive and negative categories. The dataset used in this study includes 5000 reviews, reflecting the diversity of opinions from actively participating users. The data preprocessing stage involves cleaning, case folding, tokenization, stopword removal, and lemmatization to ensure good data quality before sentiment analysis is carried out. Next, word weighting is carried out using the TF-IDF method to assign weight values to words that influence user sentiment. The research results show that the Random Forest method provides a high level of accuracy in analyzing user sentiment for Apple's App Store applications, with an accuracy of 86%, precision of 89%, recall of 81%, and f1-score of 85%. This research provides further understanding regarding user responses to Apple's App Store applications, and confirms the success of the Random Forest method in handling sentiment analysis on user review datasets on the Google Play Store.
Perbandingan Algoritma NBC, SVM dan Random Forest untuk Analisis Sentimen Implementasi Starlink pada Media Sosial X Kencono, Lintang; Darwis, Dedi
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6813

Abstract

Internet development in Indonesia continues to progress rapidly, but equitable access remains a challenge, especially in remote areas. Starlink, a satellite internet service from SpaceX, comes as a solution to reduce this gap by providing fast and stable connectivity. This research analyzes public sentiment towards the implementation of Starlink on social media platform X through a comparative approach using three Machine Learning algorithms: Naive Bayes Classifier, Support Vector Machine, and Random Forest. The research data consisted of 6,780 Indonesian tweets collected during the period September 1 to November 30, 2024 using the harvest tweet library with the keywords “starlink,” “internet starlink,” and “SpaceX starlink”. After preprocessing, a total of 5,382 tweets were used, consisting of 4,348 tweets with negative sentiment and 884 tweets with positive sentiment. To overcome data imbalance, Synthetic Minority Over-sampling Technique (SMOTE) was applied. Before the application of SMOTE, the Random Forest model showed the highest accuracy of 92%, followed by Support Vector Machine with 91%, and Naive Bayes Classifier with 85%. After SMOTE was applied, the accuracy of the three models increased significantly, with Random Forest reaching 99%, Support Vector Machine 98%, and Naive Bayes Classifier 91%. Random Forest also showed the best performance in detecting positive sentiment, with Precision and Recall values reaching 100%. This research provides an in-depth insight into the effectiveness of Machine Learning algorithms in analyzing public sentiment towards Starlink services on social media and shows that the application of SMOTE can improve the model's performance in classifying sentiment more evenly.
Pendekatan Hibrid Double Exponential Smoothing dan GRU untuk Optimasi Prediksi Harga Cabai Rawit Merah Fadila, Ika; Hartono, Budi
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6815

Abstract

Red chili is one of the staple complementary food ingredients essential to society. A rise in the price of red chili peppers can have a significant impact on the community's economy. This study develops a method combining Double Exponential Smoothing (DES) with parameter optimization through grid search and a hybrid approach using a Gated Recurrent Unit (GRU) to predict red chili prices. The goal of this approach is to improve prediction accuracy and find an appropriate solution to refine the forecasting model using double exponential smoothing. In this study, the DES method is used to capture short-term trends in historical data, while the GRU is employed to capture long-term and non-linear patterns in the data that cannot be explained by DES alone. With a data split ratio of 80% for training and 20% for testing, the lowest Mean Absolute Percentage Error (MAPE) achieved is 9.51%. This result is significantly better than using DES alone, which only yielded a MAPE of 32.74%. This study also proves to be able to improve accuracy compared to other methods, which have an average error rate of 22.32%. Therefore, this approach becomes the superior choice as a decision-support tool to anticipate extreme price increases.
Optimasi Performa Prediksi Penyakit Jantung Menggunakan Teknik Stacking Classifier Amelya, Eka; Susanto, Erliyan Redy
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6843

Abstract

Cardiovascular diseases, including heart disease, are among the leading causes of death in Indonesia. Heart disease is a condition that disrupts the function of the heart and blood vessels, often caused by blockages or narrowing of the arteries. Arteries play a crucial role in delivering oxygen-rich blood from the heart to the entire body, including the heart muscles through the coronary arteries. This condition can result from various factors such as vascular blockages, inflammation, infections, or congenital abnormalities. Such issues can impair the heart's ability to pump blood efficiently, posing a serious threat to an individual's health. This study aims to improve the accuracy of heart disease prediction by implementing the stacking classifier technique—an ensemble learning method that combines multiple machine learning algorithms, namely Support Vector Machine (SVM), Logistic Regression, and Decision Tree. The dataset used has undergone a standardization process and has been validated using the stratified k-fold cross-validation method to ensure stable predictive results. The primary contribution of this research lies in enhancing the accuracy and efficiency of heart disease diagnosis through the application of the stacking classifier, which effectively handles complex and imbalanced datasets. Previous studies have utilized the SMOTEEN technique for heart disease prediction. However, the findings of this study demonstrate that the stacking classifier approach performs better. Evaluation results show that this method achieves an accuracy of 88.52%, precision of 87.88%, recall of 90.62%, and an ROC-AUC of 94.18%, proving its effectiveness in improving medical diagnosis reliability and reducing prediction errors that could pose risks in the healthcare field.
Segmentasi Pelanggan Kartu Kredit Menggunakan Metode Klustering: Analisis dan Profiling Arifudin, Agus; Budiman, Fikri
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6879

Abstract

The use of credit cards in Indonesia has increased significantly, creating complex challenges for financial institutions in understanding user behavior and meeting their needs. This growth poses a higher risk of fraud, customer dissatisfaction due to unmet expectations, and financial instability for both consumers and banks. These issues highlight the urgency of conducting research to segment customers based on their usage behavior. The analyzed dataset includes information from 8,950 credit card users, covering transaction frequency, account balance, and transaction types. This study aims to segment customers using K-Means, DBSCAN, and Hierarchical Clustering algorithms. K-Means groups customers with similar behavioral patterns, DBSCAN identifies irregular clusters and outliers, while Hierarchical Clustering provides insights into relationships between clusters. The analysis results reveal four main segments, each with unique characteristics. For instance, the active user segment exhibits high transaction frequency and large balances, whereas new users demonstrate lower transaction frequency. These findings offer valuable insights for financial institutions to enhance their services and product offerings. By understanding the characteristics of each segment, financial institutions can tailor their marketing strategies and products to improve customer satisfaction and loyalty
Deteksi Bahan Pangan Tinggi Protein Menggunakan Model You Only Look Once (YOLO) Arjun, Restu Agil Yuli; Silmina, Esi Putri
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6889

Abstract

Stunting has a high prevalence of 21.6% from the government target of 14% and is one of the health problems in Indonesia. Lack of nutrition, especially protein, is the main cause that plays a role in child growth. One of the preventive solutions is to provide protein-rich complementary foods (MP-ASI). To enhance this solution, technology that can swiftly and precisely identify high-protein food components is imperative. This research seeks to create a high-protein food detection model utilizing the YOLOv11 framework, chosen for its efficacy in object detection, particularly in intricate environments and with overlapping items. The research methodology includes several stages: dataset collection and annotation, data pre-processing, model training, model evaluation, and model testing. The dataset is divided into three parts: 70% for the training set, 20% for the validation set, and 10% for the test set. The YOLOv11s model is used for training. Evaluation is based on precision, recall, and mean Average Precision (mAP) metrics to ensure the model’s detection accuracy. The evaluation results indicate a precision of 96%, recall of 92.3%, mAP50 of 96.4%, and mAP50-95 of 81.5%. During testing, the model achieved a success rate of 98.2%. These results demonstrate the model’s potential in detecting protein-rich foods, which could significantly contribute to addressing malnutrition and stunting.
Optimasi Analisis Sentimen Twitter Tentang Isu Kesehatan Mental dengan Bi-LSTM pada Dataset Tidak Berimbang Fatmawati, Indah Rani; Putra, Muhammad Pajar Kharisma
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6890

Abstract

This study aims to analyze Twitter user sentiments related to mental health issues using the Bidirectional Long Short-Term Memory (BiLSTM) model. The dataset consists of 52,681 entries covering seven mental health categories: Anxiety, Bipolar, Depression, Normal, Personality Disorder, Stress, and Suicidal. The methods used include data pre-processing, data splitting, and model training with class weight adjustment techniques to handle data imbalance. The training results show an increase in accuracy from 16.02% in the first epoch to 88.48% in the 10th epoch, with an evaluation accuracy of 74.21%. The model shows the best performance in the Anxiety class with an F1-score of 0.90. However, the model still experiences limitations in classifying minority classes such as Bipolar and Personality Disorder due to the small amount of data and the complexity of language expressions in these categories. Therefore, an increase in the amount of data and more adaptive language processing techniques are needed to improve model performance in categories with limited data.
Perbandingan Algoritma Naïve Bayes dan Random Forest untuk Melakukan Analisis Sentimen Cyberbullying Generasi Z Pada Twitter Danuarta, Ervin; Alita, Debby
Building of Informatics, Technology and Science (BITS) Vol 6 No 4 (2025): March 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v6i4.6909

Abstract

Cyberbullying is a significant social problem, especially for Generation Z,who actively use social media such as Twitter, Instagram and TikTok. It has a very negative impact on the victim's mental health, such as a sense of isolation, loss of confidence, and insecurity. This study aims to compare the performance of two machine learning algorithms, namely Naive Bayes and Random Forest, in sentiment analysis related to cyberbullying in Generation Z through the Twitter platform. The research method involved collecting and preprocessing data from 5505 tweets, which were then divided into training data (80%) and test data (20%). The research also applied Synthetic Minority Oversampling Technique (SMOTE) to overcome data imbalance. Preliminary results show that before the application of SMOTE, Naïve Bayes had an accuracy of 92% and Random Forest reached 94%. After the application of SMOTE, the performance of both algorithms changed. Naive Bayes accuracy decreased to 89%, with precision increasing from 92% to 99% for negative sentiments, but recall dropped from 100% to 79%, resulting in an F1-Score of 88%. In contrast, Random Forest showed significant improvement, with accuracy reaching 100%, precision and recall for negative sentiment remaining 100%, and F1-Score increasing from 97% to 100%. This study concludes that Random Forest, with the application of SMOTE, provides more stable and effective performance than Naive Bayes in cyberbullying sentiment analysis. These results are expected to support the development of text analysis technology and efforts to prevent cyberbullying in Generation Z.