Jurnal Nasional Teknik Elektro dan Teknologi Informasi

Comparative Analysis of MVVM and MVP Patterns Performance on Android Dashboard System

Fajar Pradana — 2025-05-16

The rapid growth of the Android market in various developing countries has driven the demand for higher-quality applications. Developing Android-based applications presents specific challenges, such as the need for responsive designs and optimization for devices with diverse specifications. Design patterns like model-view-controller (MVC), model-view-presenter (MVP), and model-view-viewmodel (MVVM) have become popular approaches to address these issues. However, studies on the performance of design patterns in Android applications, especially in modern programming languages like Kotlin, remain limited. This research aims to compare the performance of the MVP and MVVM design patterns in an Android-based boarding house management application, KosGX. This application utilized Kotlin and featured an interactive dashboard requiring significant device resources. Testing was conducted by measuring performance across three key aspects: central processing unit (CPU) usage, memory usage, and system response time. The results of the study showed that MVVM outperformed in CPU efficiency, with an average usage of 8.92% compared to 11.15% for MVP. In terms of memory usage, MVVM was also slightly more efficient, with an average usage of 121.48 MB compared to 121.55 MB for MVP. However, MVP excelled in response time, averaging 236.88 ms, whereas MVVM reached 252.68 ms. This study underscores that the choice of design pattern affects application performance. MVVM is more efficient in CPU and memory usage, while MVP offers better response times. These findings provide valuable insights for developers in selecting the optimal design pattern based on the specific needs of their applications.

Detecting Fraudulent Transaction in Banking Sector Using Rule-Based Model and Machine Learning

Cut Dinda Rizki Amirillah — 2025-05-27

This research aims to develop an effective fraud detection model in banking transactions using the rule-based model (RBM) approach and the isolation forest (IF) machine learning algorithm. Based on data from the Ministry of Communication and Information Technology, there were more than 405,000 online fraud cases during the 2019–2022 period, indicating the need for a reliable fraud detection system to protect customers. The research method involves collecting banking transaction data for four months through channels such as ATM, internet banking, and mobile banking. The RBM model was used as an initial approach, detecting suspicious transaction patterns based on defined rules. However, it has limitations in detecting transactions that are not defined in the rules. To complement this shortcoming, this research implemented IF, an effective unsupervised learning model for detecting anomalies using the isolation tree (iTree) technique to identify suspicious transactions. The results showed that the IF model could detect anomalous patterns not covered by RBM, thereby improving the accuracy of fraud transaction identification. The precision data of 99% indicates that the model’s predictions of anomalies are indeed anomalies, while a recall value of 1.0 shows that the model successfully identified all anomalies in the dataset. In conclusion, the combination of RBM and IF provides a comprehensive approach to fraud detection in the banking sector. IF’s ability to detect anomalies more dynamically and accurately can reduce fraud losses in the industry.

Advancing Indonesia’s Energy Transition: Coal-to-Nuclear (C2N) Simulation Study

Irfan Eko Budiyanto — 2025-05-28

Indonesia’s ambitious net-zero emissions (NZE) target for 2060 necessitates a transformative shift from coal-dependent power systems to nuclear energy sources. Although previous studies have explored various aspects of energy transitions, limited attention has focused on the possibility of replacing existing coal-fired boilers with nuclear reactors while maintaining the input and output parameters of the current infrastructure. This paper presents an approach to facilitating Indonesia’s nuclear energy transition by a coal-to-nuclear (C2N) simulation study. Specifically, the focus is on the 400 MW coal-fired power plant (CFPP), which relies on coal combustion. This study was designed to model the conversion process of the coal combustion system to a modular nuclear reactor setup while preserving the existing steam turbine and generator infrastructure. The foundation of this study was that the replacement of the nuclear reactor-based heat source must meet the requirements of the pre-existing water and steam cycle design. Various configurations for substituting the current boiler with a nuclear reactor were analyzed in this study, by considering engineering, operational, and modeling aspects. Results from model simulations for nine different operating conditions showed a deviation of the main-steam temperature of about 3% of the design value, starting at 120 MWe and above. Nevertheless, all other parameters of the conversion simulation results demonstrated a very small deviation. The deviation was close to the actual existing operational conditions of the previous CFPP. This paper highlights how the simulation demonstrated a promising integration of legacy infrastructure with emerging nuclear technology.

Semantic Similarity Measurement Evaluation of KBBI Synonyms Using a Word Embedding Approac

Muhammad Rafli Aditya H. — 2025-05-28

Kamus Besar Bahasa Indonesia (KBBI) is a primary resource for data in research on determining word-meaning similarity in Indonesian. This study investigates the effectiveness of word embedding methods and the term frequency–inverse document frequency (TF-IDF) weighting technique in assessing the semantic similarity of synonym pairs. The objective is to measure the similarity of synonym word pairs listed in KBBI by applying cosine similarity, leveraging TF-IDF weighting, various word embedding models, and latent semantic analysis (LSA). The methodology involved data collection, followed by a text preprocessing stage consisting of case folding, stopword removal, stemming, and tokenization. The processed data were transformed into vector representations using word embedding models, including Word2Vec, fastText, GloVe, and sentence-bidirectional encoder representations from transformers (S-BERT), and TF-IDF. LSA was employed for dimensionality reduction of the vectors before similarity testing using cosine similarity, with final evaluation of the results. The findings revealed that fastText significantly improved the similarity scores between synonym pairs, achieving an average similarity score of 0.901 for 30 synonym pairs. Evaluation results indicated an accuracy of 0.88, a recall of 1.00, a precision of 0.81, and an F1 score of 0.90. These results suggest that fastText is more effective in enhancing the accuracy of synonym meaning similarity measurements. Future research is encouraged to expand the corpus and further explore the use of word embedding for semantic similarity tasks. This study contributes to the natural language processing advancement and provides a potential foundation for more accurate language-based applications that assess word meaning similarity in KBBI.

5G Network Planning Using Macrocell and Piccocell Technology

Rivan Achmad Nugroho — 2025-05-28

The city of Pontianak is projected to experience substantial growth in network demand, driven by the expansion of commercial hubs, educational institutions, tourism destinations, and essential public services. Under the current status quo, Pontianak lacks 5G network coverage, underscoring the necessity of implementing a comprehensive 5G network plan to support its urban development. This research conducted a detailed analysis of 5G network coverage and capacity planning, utilizing macrocell and picocell technologies to address the connectivity demands of an urban environment. Operating within the 3.5 GHz frequency band with a 100 MHz bandwidth, this research examined network requirements in the medium band spectrum. The results revealed that macrocell technology required 18 uplink and 23 downlink sites to cover an area of 107.8 km², while picocell technology demanded a denser infrastructure, comprising 351 uplink and 364 downlink sites across 90.72 km². Based on a five-year capacity projection for a population of 673,400, the macrocell technology will require 10 uplink and 22 downlink sites. On the contrary, picocell technology, which is more suitable for densely populated areas, will require 261 uplink and 263 downlink sites to serve a population of 423,881. Simulation results indicated that synchronization signal reference signal received power (SS-RSRP) and secondary synchronization signal received power (SS-SINR) values met or surpassed the established key performance indicators (KPI) for both technologies. This 5G network plan aligns with Pontianak’s smart city vision by enhancing connectivity, optimizing coverage, and delivering seamless user experiences, highlighting the adaptability of macrocell and picocell solutions in varied urban settings.

Comparison Study of Object-Relational Mapping Performance Based on the Implementation of the DSAP

Muhammad Rezy Anshari — 2025-05-28

Object-relational mapping (ORM) is a technique that maps in-memory objects and tables in the database, implementing data source architectural patterns (DSAP), namely Data Mapper and Active Record. These patterns require comparison due to performance difference indications and their significant roles in a system’s business processes. This study aims to compare and analyze execution duration and memory consumption quantitatively, and functions influencing them in the ORM, utilizing Data Mapper and Active Record. The objects were Doctrine (Data Mapper) and Eloquent (Active Record). The performance profiling in the ORM was conducted as a library rather than a framework. This profiling encompassed create, read, update, and delete (CRUD) and lookup operations based on specified measurement metrics, conducted using variations in the number of database records. The profiling process was automated using a script, leveraging a combination of Xdebug and Apache Benchmark. The analysis employed using Kcachegrind and big O notation, resulting in performance graphics, relative percentage differences, and functions’ contributions to the performance. Results showed that memory consumption outperformed Data Mapper. Data Mapper was superior in execution duration in most operation combinations and metrics. Function groups of database transactions, object serialization, and retrieval records were the primary contributors to the performance. Object and database synchronizations became additional contributors to Active Record. The complexity of the largest contributor functions in Data Mapper was higher than that of Active Record. Future studies can utilize automation concepts in the profiling process and substitute Xdebug according to the requirements of the programming languages used by the ORM.

Deteksi Duplikasi Data pada Sistem Pemantauan Kualitas Udara Berbasis IoT

Dwi Ilham Maulana — 2025-05-28

The increasing volume of data on the Internet of things (IoT)-based systems has driven the need for efficiency in data management, particularly in air quality monitoring systems. One approach to address this challenge is data duplication detection, which works to eliminate redundant data to reduce storage requirements and power consumption. This study aims to develop an IoT-based air quality monitoring system incorporating a data duplication detection method as part of an effort to support the green IoT concept. The methodology involved a comparative analysis between systems with and without the implementation of data duplication detection, accompanied by a comprehensive evaluation of system performance. The data tested included the size of transmitted data and device power consumption during the transmission process. Testing was conducted under real operational conditions over a 24-hour period. The results indicate that the implementation of data duplication detection successfully reduced the size of transmitted data from 56 bytes to 11–44 bytes, depending on the level of data redundancy. Power consumption was reduced by 1.59% to 3.84% compared to the system without data duplication detection. This method was also proven not to affect the accuracy of the displayed data, thereby maintaining the system’s functional requirements. In conclusion, the implementation of the data duplication detection method in an IoT-based air quality monitoring system not only optimizes data transmission processes but also supports energy efficiency in line with the principles of green IoT. This research provides a significant contribution to the development of more sustainable and energy-efficient IoT systems.

Ensemble Voting Classifier Berbasis Multi-Algoritma dan Metode SMOTE untuk Klasifikasi Penyakit Jantung

Dede Kurniadi — 2025-05-28

Jantung adalah organ penting tubuh yang berfungsi untuk memompa darah. Gangguan pada jantung dapat mengganggu sirkulasi darah dalam tubuh dan menjadi salah satu penyebab utama kematian global. Menurut laporan World Health Organization (WHO) tahun 2021, jumlah kematian akibat penyakit jantung mencapai angka signifikan. Sementara itu, prevalensi penyakit jantung di Indonesia mencapai 1,5%. Maka, diperlukan upaya pencegahan dan deteksi dini penyakit jantung dengan memanfaatkan teknologi pembelajaran mesin. Penelitian ini bertujuan untuk mengembangkan model klasifikasi penyakit jantung menggunakan algoritma naïve Bayes dan random forest melalui pendekatan ensemble voting classifier. Data yang digunakan berasal dari Kaggle, yang terdiri atas 1.000 record dengan 14 variabel, satu di antaranya sebagai target klasifikasi. Ketidakseimbangan data diatasi dengan teknik synthetic minority oversampling technique (SMOTE), sedangkan seleksi fitur dikonsultasikan dengan dokter spesialis jantung untuk memastikan relevansi klinis. Model dilatih menggunakan algoritma naïve Bayes, random forest, serta kombinasi keduanya melalui metode ensemble voting classifier, berbeda dengan penelitian sebelumnya yang hanya membandingkan beberapa algoritma untuk menentukan akurasi tertinggi. Hasil pengujian menunjukkan bahwa model yang dilatih dengan ensemble voting classifier memiliki kinerja terbaik, dengan akurasi 98,28%, presisi 98,41%, recall 98,41%, dan F1-score 98,41%. Penelitian ini membuktikan bahwa penerapan metode ensemble voting classifier mampu mencapai akurasi yang lebih baik dibandingkan penggunaan algoritma secara terpisah. Model ini termasuk kategori excellent classification dan diharapkan dapat berkontribusi dalam bidang kedokteran serta mendukung praktisi medis dalam pengembangan sistem pendukung keputusan untuk diagnosis penyakit jantung.

Developing a Drowsiness Detection System for Safe Driving Using YOLOv9

Fernando Candra Yulianto — 2025-05-28

Drowsiness detection systems play a crucial role in safe driving, considering the high rate of traffic accidents caused mainly by drowsiness. Several drowsiness detection systems built using the eye aspect ratio (EAR), percentage of eyelid closure (PERCLOS), and convolutional neural network (CNN) methods still have limitations in terms of accuracy and response time. This study aimed to overcome these problems by applying the You Only Look Once version 9 (YOLOv9). This method has advantages in terms of speed and accuracy because it can detect objects in real-time in one processing stage. The dataset was collected independently from several sources in a real environment inside the vehicle with various lighting and viewing angles; then, labeling, preprocessing, and modeling processes were conducted. The model performance was evaluated based on precision, recall, F1 score, and mean average precision (mAP) metrics. The best model was optimized using several optimization techniques to determine the most optimal results. The results indicate that the YOLOv9 model trained using Nadam (Nesterov-accelerated adaptive moment estimation) optimization has a better image processing speed than other models. This model yielded a precision level of 99.4%, recall of 99.6%, F1 score of 99.5%, mAP@50 of 99.5%, mAP@50-95 of 85.5%, and a processing speed of 52.08 FPS. The developed model can detect drivers’ drowsiness signs, such as closed eyes, yawning, abnormal head positions, and unnatural hand movements, in real time. However, this model still has limitations in detecting drivers wearing sunglasses, so further development is needed to improve its performance in these conditions.

Deteksi Berita Hoaks Berbahasa Indonesia Menggunakan 1-Dimensional Convolutional Neural Network

Muhammad Zuama Al Amin — 2025-05-28

Kemajuan teknologi informasi yang pesat telah memfasilitasi penyebaran informasi secara global, tetapi juga meningkatkan prevalensi berita hoaks, khususnya di Indonesia. Berita hoaks memiliki potensi untuk menciptakan disinformasi yang dapat memengaruhi opini publik, stabilitas sosial, dan keamanan. Oleh karena itu, diperlukan solusi berbasis teknologi yang efektif untuk mendeteksi dan mengidentifikasi hoaks atau berita palsu. Penelitian ini bertujuan untuk mengembangkan dan mengoptimalkan model 1-dimentional convolutional neural network (1D CNN) guna mendeteksi berita hoaks dengan tingkat akurasi yang tinggi. Dataset yang digunakan terdiri atas 12.151 artikel, yang mencakup 5.276 berita valid dan 6.875 berita hoaks, yang diperoleh dari sumber tepercaya serta platform antihoaks. Tahapan prapemrosesan teks meliputi pembersihan data, case folding, penghapusan tanda baca, penghapusan angka, dan penghapusan stopword. Data tekstual selanjutnya diproses melalui tahapan tokenisasi dan padding untuk persiapan pelatihan model. Arsitektur 1D-CNN yang diusulkan mengintegrasikan lapisan embedding, conv1d, batch normalization, globalmaxpooling1d, dense, dan dropout layer, untuk meningkatkan kemampuan generalisasi serta mengurangi risiko overfitting. Model dilatih menggunakan optimizer Adam dan evaluasi kinerja menggunakan 10-fold cross-validation. Hasil eksperimen menunjukkan bahwa model menghasilkan rata-rata akurasi 97,74%, presisi 97,75%, recall 97,74%, dan F1-score 97,73%. Model yang dikembangkan mampu mengungguli metode sebelumnya, seperti kombinasi convolutional neural network- bidirectional long short – term memory neural network (CNN-BiLSTM), gated recurrent unit (GRU), dan metode konvensional seperti naïve Bayes atau support vector machine (SVM), baik dari segi akurasi maupun efisiensi waktu pelatihan. Penelitian ini menunjukkan bahwa model memiliki kemampuan yang andal dalam mengidentifikasi berita hoaks, baik dari segi tingkat deteksi yang akurat maupun konsistensi kinerja.