cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
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
Published by Forum Kerjasama Pendidikan TInggi
ISSN : 26848910     EISSN : 26853310     DOI : -
Core Subject : Science,
Computer Science & IT
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
 75   76   77   78   79 
Implementasi Convolutional Neural Network dengan SMOTE+ENN untuk Klasifikasi Kualitas Udara Berdasarkan Data Deret Waktu Polutan Santoso, Cahyono Budy; Kesya Makarena, Maria Rachel
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8057

Abstract

The degradation of air quality in metropolitan areas, such as Jakarta, constitutes a significant environmental and public health challenge, contributing directly to an elevated risk of various diseases. The primary objective of this study is to develop and evaluate the effectiveness of an air quality classification model based on a Convolutional Neural Network (CNN), with a specific focus on addressing class imbalance using the hybrid resampling technique SMOTE+ENN. Utilizing a historical dataset from the HI Jakarta Station spanning 2010-2021, the model leverages key pollutant parameters (PM10, SO₂, CO, O₃, and NO₂) to classify air quality according to the Indonesian Air Quality Index (ISPU) standard. To mitigate the inherent challenge of class imbalance within the dataset, this study conducts a comparative analysis between a baseline CNN model and an optimized model enhanced with the hybrid resampling technique, Synthetic Minority Over-sampling Technique and Edited Nearest Neighbours (SMOTE + ENN). The dataset was partitioned into an 80% training set and a 20% testing set. Empirical results demonstrate that the application of SMOTE + ENN yields a substantial improvement in performance. The final optimized model achieves a superior accuracy of 98.98%, significantly outperforming the baseline model. This outcome confirms that integrating CNN with the SMOTE + ENN strategy produces a highly effective and robust framework for air quality classification in Jakarta. Nonetheless, subsequent validation on more diverse datasets is recommended to ascertain the model's generalization capabilities and long-term reliability.
Sentiment Classification and Interpretation of Tokopedia Reviews: A Machine Learning, IndoBERT, and LIME Approach Mbake Woka, Adrian Yoris; Purbolaksono, Mahendra Dwifebri; Utama, Dody Qori
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8072

Abstract

Sentiment classification of user reviews plays a vital role in business decision-making, especially on e-commerce platforms like Tokopedia. This study evaluates the performance of various sentiment classification models such as Logistic Regression LinearSVC, and BERT models, both baseline and fine-tuned. Evaluation metrics used include accuracy, precision, recall, and F1-score, applied to Tokopedia review data labelled based on user ratings. The result is fine-tuned BERT model has the best and consistent result, with 92% accuracy and 0.92 f1-score for each class. This shows that fine-tuned BERT can effectively capture the semantic context of user reviews. Its consistent performance across classes makes it suitable for reliable sentiment classification in real-world applications. Furthermore, fine-tune BERT model is visualized by Local Interpretable Model-agnostic Explanation to identify features – in this case is word – that indicates sentiment as positive or negative. It will show as color, orange for positive and blue as negative. This method will make the model more transparent and more reliable.
Sentiment Analysis of ChatGPT App Reviews on the Play Store Using KNN and Decision Tree Methods Rahman, Hammam Aulia Nur
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8078

Abstract

This study aims to analyze the sentiment of user reviews of the ChatGPT application on the Google Play Store, a platform that directly reflects public opinion toward this increasingly popular artificial intelligence application. A total of 10,000 reviews were collected through web scraping and underwent a series of rigorous preprocessing stages. These stages included data cleaning to remove noise, case folding to standardize text, tokenizing to break sentences into words, normalization to standardize informal words, and stopword removal to eliminate common but uninformative words—ensuring optimal data quality. Feature weighting was then performed using the Term Frequency-Inverse Document Frequency (TF-IDF) method with three n-gram scenarios (Unigram, Unigram+Bigram, Unigram+Trigram), followed by feature selection using Chi-Square to identify the most relevant features. The processed and weighted data were then classified using two machine learning algorithms: K-Nearest Neighbors (KNN) and Decision Tree. The evaluation results show that the Decision Tree model with Unigram+Bigram features achieved the highest accuracy of 0.8089 (80.89%) and an F1-Score of 0.8894 (88.94%), making it the best-performing model in this study. These findings provide valuable insights for application developers to better understand user perceptions, identify areas for improvement, and enhance the quality of ChatGPT services in the future, especially when addressing the challenge of imbalanced review data.
Pendekatan LSTM Berbasis Deep Learning dalam Memprediksi Fluktuasi Harga Cabai Pertiwi, Aryka Anisa; Harani, Nisa Hanum; Prianto, Cahyo
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8100

Abstract

The significant fluctuation in chili prices in Indonesia leads to economic instability, particularly for consumers and market stakeholders. This study aims to develop a daily chili price prediction model using the Long Short-Term Memory (LSTM) algorithm based on deep learning, designed to capture seasonal patterns and long-term dependencies in historical data. The research adopts the CRISP-DM approach, encompassing business understanding, data processing, model training, and implementation into a web-based dashboard. The dataset, collected from Pagar Alam City between 2022 and 2024, includes features such as previous prices, chili sub-variants, sinusoidal time transformations, and market conditions. The LSTM regression model demonstrated high performance, achieving an R² score of 0.9567, a MAE of 1,402.92, and an RMSE of 2,595.98. Additionally, a classification model was developed to predict price status (increase, decrease, stable) as a decision-support tool. The deployment of this system into an interactive dashboard enables real-time price predictions. These results indicate that the LSTM-based approach is not only technically accurate but also offers a practical solution for commodity price monitoring and decision-making in the food sector.
Analisis Perbandingan Algoritma K-Means dan K-Medoids untuk Pengelompokan Sentimen Ulasan Aplikasi E-Commerce Pecaro, R Immanuel Giovanni Italiano; Setiaji, Galet Guntoro; Rifa'i, Ahmad
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8145

Abstract

The growing digital technology has driven the rapid growth of E-Commerce applications, which is characterized by the number of similar applications available on the Google Play Store. This phenomenon has led to an increase in people's need for convenience in online shopping, as well as showing increasingly fierce competition in the digital marketplace. Reviews on E-Commerce applications in the Google Play Store often serve as a basis for users to decide whether to download an application. These reviews provide valuable insights, allowing users to assess whether the application is worth downloading. Shopee, one of the largest E-Commerce applications in Indonesia, currently has more than 15 million ratings and reviews on the Google Play Store. This study aims to compare the performance of the K-Means and K-Medoids algorithms in clustering numerical data from application reviews. Clustering was performed using the Clustering technique based on two numerical variables, namely score and thumbsupcount, to provide an initial overview of user opinion trends regarding the application. The dataset, consisting of 500 reviews, was collected from the Google Play Store in December 2024. The results Davies-Bouldin Index of the study indicate that K-Means outperforms K-Medoids, with a comparison score of 0.457 to 0.803.
Deteksi Dini Depresi Mahasiswa Tingkat Akhir Menggunakan Algoritma Naïve Bayes dan Instrumen PHQ-9 Sadikin, Muhammad; Dwiki Putri, Dini Ridha; Amanda, Azrifirizky
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8162

Abstract

This study aims to analyze the performance of the Naïve Bayes algorithm in an early detection system for depression levels among final-year university students using the PHQ-9 instrument. The classification is conducted in a multi-category manner, grouping respondents into normal, mild depression, moderate depression, and severe depression categories based on the results of the PHQ-9 questionnaire. The main issues addressed in this research are the low awareness of depression symptoms and the need for a screening tool that is fast, lightweight, and accessible. Involving 188 respondents, data were collected through the PHQ-9 questionnaire and subsequently processed using the Naïve Bayes method. Model evaluation was performed using a confusion matrix and 10-fold cross-validation, resulting in an accuracy of 87.23%, a weighted average precision of 87.54%, and a weighted average recall of 87.25%. These findings demonstrate that Naïve Bayes can classify depression levels with high and stable accuracy, particularly in the moderate and severe depression categories. This study recommends the use of the Naïve Bayes algorithm as the basis for developing a web-based screening system that can be utilized by higher education institutions as a self-assessment and systematic early detection tool for depression disorders.
Analisis Model Klasifikasi Sentimen Publik Terhadap Kebijakan Keberlanjutan IKN Menggunakan BERT Sebagai Feature Extractor dan K-Nearest Neighbor (KNN) Fiqri, Mohammad Hiqmal; Rudiman, Rudiman; Verdikha, Naufal Azmi
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8168

Abstract

This study aims to evaluate the performance of sentiment classification models for public opinions regarding the relocation of Indonesia’s new capital (IKN) using a combination of IndoBERT as a feature extractor and K-Nearest Neighbor (KNN) as a classifier. The dataset consisted of 1,274 YouTube comments related to IKN, which were annotated by an expert in sociology and text analysis. The preprocessing stage involved cleaning numbers, URLs, emojis, and punctuation, as well as removing stopwords using the Sastrawi library. IndoBERT produced 768-dimensional vector representations, which were then classified using KNN with k=5 and Euclidean distance. Evaluation with 5-fold cross validation achieved an accuracy of 73.31%. However, the recall for the positive class was relatively low (0.49), indicating challenges in detecting positive comments due to class imbalance (831 negative, 294 positive, 149 neutral). These findings suggest that the IndoBERT+KNN model performs well on majority classes but struggles with minority classes. The contribution of this research is to provide a critical analysis of the limitations of IndoBERT-based models in Indonesian sentiment classification and to recommend future directions, including data balancing and fine-tuning approaches.
Implementasi Grid Search CV KNN dengan Preprocessing Z-Score Outlier Removal untuk Sistem Prediksi Risiko Kehamilan Anggita, Ivan Maulana; Naufal, Muhammad; Zami, Farrikh Al
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8206

Abstract

This study aims to optimize the K-Nearest Neighbors (KNN) algorithm in predicting pregnancy risk levels using the “maternal health risk” dataset from the UCI Machine Learning Repository. The methodology includes data preprocessing through outlier detection and removal using Z-score, normalization with Standard Scaling, and categorical encoding on the target labels. Hyperparameter tuning is performed using GridSearchCV to identify the optimal combination of KNN parameters (number of neighbors, distance weight, and distance metric). The results show that the unoptimized KNN model achieved an accuracy of only 69.46%, whereas the optimized model reached an accuracy of 82.00%, with macro average precision of 81.91%, recall of 82.89%, and F1-score of 82.23%. Evaluation using a confusion matrix also revealed significant performance improvement, especially in the high-risk category. The optimized model was deployed as a web application using the Flask framework and Docker via Hugging Face Spaces, enabling real-time and efficient online pregnancy prediction. These findings indicate that combining KNN with GridSearchCV and data normalization significantly enhances prediction performance and offers practical application in healthcare decision support systems.
Optimalisasi Arsitektur LSTM dengan Pendekatan Bidirectional untuk Deteksi Kantuk Pengemudi Berbasis Fitur Wajah Hartono, Andhika Rhaifahrizal; Naufal, Muhammad; Alzami, Farrikh
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8219

Abstract

Traffic accidents caused by driver fatigue and drowsiness remain a serious safety concern in many countries, including Indonesia. Various image-based drowsiness detection systems have been developed, yet many still rely on single-frame analysis and lack the ability to capture complete temporal context. To address this issue, a system capable of accurately and real-time detecting signs of drowsiness is required. This study aims to evaluate and compare the performance of Long Short-Term Memory (LSTM) and Bidirectional LSTM (BiLSTM) algorithms for a facial-feature-based drowsiness detection system. The dataset used is YawDD, which consists of videos of drivers yawning and in neutral conditions. Each video was decomposed into frames and analyzed using MediaPipe to extract facial landmarks. Two main features, Eye Aspect Ratio (EAR) and Mouth Opening Ratio (MOR), were utilized. Due to class imbalance, the SMOTE technique was applied to the minority class in the training data. Both LSTM and BiLSTM models were compared under similar architecture configurations. The results show that BiLSTM outperformed LSTM with an accuracy of 94,74% and an F1- score 94,82%, compared to 92,98% accuracy and 93,22% F1-score achieved by LSTM. These findings demonstrate that bidirectional sequential processing in BiLSTM is more effective in capturing the temporal patterns of drowsiness symptoms. This study contributes to the development of accurate and efficient computer vision-based drowsiness detection systems.
Optimalisasi Model CNN dengan Teknik Kontras Lokal CLAHE untuk Klasifikasi Pneumonia pada Citra X-Ray Salma, Rania Alfita; Aditya, Christian Sri Kusuma
Building of Informatics, Technology and Science (BITS) Vol 7 No 2 (2025): September 2025
Publisher : Forum Kerjasama Pendidikan Tinggi

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bits.v7i2.8271

Abstract

Pneumonia is a lung infection that has a widespread impact on public health, particularly in areas with limited access to healthcare services. Chest X-ray imaging plays an important role in diagnosing this disease; however, low contrast quality often becomes an obstacle to automated classification using deep learning methods. This study aims to evaluate the effectiveness of the Contrast Limited Adaptive Histogram Equalization (CLAHE) method in enhancing the visual quality of chest X-ray images and to analyze its impact on the performance of a Convolutional Neural Network (CNN) model in detecting pneumonia. CLAHE enhances the visibility of radiographic details through local contrast redistribution with a clip limit, allowing previously indistinct pathological structures to be more clearly recognized by the CNN. The dataset used consists of 2,623 X-ray images that are divided into two classes, namely Normal and Pneumonia. The training process was conducted under two scenarios, without and with the application of CLAHE. The evaluation results show that the CNN model without CLAHE achieved an accuracy of 96.18%, while the model with CLAHE improved to 99.69%. This improvement is significant as it reduced the classification error rate from approximately 3.8% in the model without CLAHE to only 0.3% in the model with CLAHE, while also increasing precision, recall, and f1-score across all classes. Therefore, combination of CLAHE and CNN can be applied as an effective approach for pneumonia detection that is accurate, consistent, and efficient, especially in environments with limited computational resources.

Page 77 of 93 | Total Record : 926

 75   76   77   78   79 

Filter by Year

2019 2025


Reset
Filter By Issues
All Issue Vol 7 No 3 (2025): December 2025 Vol 7 No 2 (2025): September 2025 Vol 7 No 1 (2025): June (2025) Vol 6 No 4 (2025): March 2025 Vol 6 No 3 (2024): December 2024 Vol 6 No 2 (2024): September 2024 Vol 6 No 1 (2024): June 2024 Vol 5 No 4 (2024): March 2024 Vol 5 No 3 (2023): December 2023 Vol 5 No 2 (2023): September 2023 Vol 5 No 1 (2023): June 2023 Vol 4 No 4 (2023): March 2023 Vol 4 No 3 (2022): December 2022 Vol 4 No 2 (2022): September 2022 Vol 4 No 1 (2022): June 2022 Vol 3 No 4 (2022): March 2022 Vol 3 No 3 (2021): December 2021 Vol 3 No 2 (2021): September 2021 Vol 3 No 1 (2021): June 2021 Vol 2 No 2 (2020): December 2020 Vol 2 No 1 (2020): June 2020 Vol 1 No 2 (2019): December 2019 Vol 1 No 1 (2019): June 2019 Vol 1 No 1 (2019): Edisi Juni More Issue