Journal of System and Computer Engineering

A Web-Based Teacher Performance Behavior Evaluation Using BARS Method

Farhan Pratama — 2026-01-29

The assessment of teachers’ professional performance behavior plays a crucial role in improving the quality if educatuin. However, at Sekolah Dasar Negeri 01 Sungai Raya Kepulauan, the assessment system still relies on MS Excel and conventional document filing, leading to limitations in transparency, efficiency, and accuracy. Teachers and school principals face difficulties in managing assessment data, making the process suboptimal. To address this issue, a web-based assessment system was developed using the Behaviorally Anchored Rating Scale (BARS) method, which aligns with ASN BerAKHLAK values. This system allows for flexible teacher performance behavior assessments via electronic devices such as smartphones and computers. The system development method employed is the Systems Development Life Cycle (SDLC) with the Rapid Application Development (RAD) model. The system is implemented using PHP as the programming language, MySQL as the database, and designed with Unified Modeling Language (UML). Black Box tesing result indicated that the system successfully meets user needs and enhaces the efficiency and transparency of teacher performance behavior assessments in accordance with ASN BerAKHLAK values.

Performance Comparison of Ion Lithium Batteries and Lead Acid Batteries in Electrical Energy Storage Systems

musrawati ST M.Si — 2026-01-29

Enrekang Regency, which is mostly mountainous and has an increasing need for a reliable electrical energy storage system along with the development of renewable energy technologies such as solar panels and wind turbines. Two types of batteries commonly used are ion lithium batteries and lead acid batteries. Research Objectives To determine the effectiveness of the performance of ion lithium batteries and lead acid batteries and to determine the relative energy storage capacity of these two types of batteries and to determine whether one of them has an advantage in greater storage capacity. The method used in this study is This study uses a comparative descriptive approach with a literature study method (library research) and quantitative data analysis. The aim is to compare the technical performance of two types of batteries based on relevant and valid secondary data. The results of this study indicate that ion lithium batteries have the advantage of lasting a long time when discharged with a load, ion lithium batteries last for 8 hours 43 minutes, compared to lead acid which only lasts for 7 hours 48 minutes, this data collection is carried out by charging and discharging tests, life cycle tests, self-tests, and safety tests

Prediction of Protein Content of Shredded Goldfish Based on Physical Characteristics and Processing Process Using Random Forest Regression Method

Irene Devi Damayanti — 2026-01-29

Shredded goldfish is a processed fishery product that has high nutritional value, especially in its protein content. This study aims to predict the protein content in shredded goldfish based on the physical characteristics of the ingredients (moisture, ash, fat, and crude fiber content) and processing parameters (temperature and frying time) using the Random Forest method. The data used consisted of 10 samples of proximate analysis results and were divided into training data (67%) and test data (33%). The model was evaluated using MAE, MSE, RMSE, and R-squared metrics. The evaluation results showed that the model produced an MAE of 0.5649, MSE of 0.5409, RMSE of 0.7354, and R² of 0.0898. The low R² value indicates that the model is still not optimal in explaining variations in the target data. The prediction of protein levels for new data with certain characteristics resulted in a value of 20.16%, which is still within the range of actual values. This research shows the potential of using machine learning methods in predicting the nutritional value of food products, although increased accuracy is still needed through additional data and exploration of other models. It is recommended that the frying temperature is 155°C to 160°C and the frying time is 11 minutes to 13 minutes to maintain optimal protein levels.

Design of IoT-Based Energy Meter for Efficiency and Disturbance Detection

Bayu Adrian Ashad Bayu — 2026-01-29

The increasing need for energy consumption monitoring has driven the development of systems capable of providing accurate electrical information and detecting disturbances at an early stage. This study aims to design an IoT-Based Energy Meter capable of monitoring electrical parameters in real time and detecting load anomalies as a basis for energy efficiency analysis. The system uses a PZEM-004T sensor and an ESP32 microcontroller to measure voltage, current, power, energy, and power factor (cos φ). The data is transmitted to an IoT platform via a wireless connection so it can be monitored remotely. A Long Short-Term Memory (LSTM) model is applied to identify normal power consumption patterns and detect deviations, while a rule-based method is used to detect critical conditions such as overcurrent. Test results show that the device is capable of performing measurements with high accuracy, with error percentages for voltage, current, power, and cos φ parameters ranging between 0%–5% for three types of loads: iron, electric fan, and refrigerator. The LSTM model also successfully detects anomalies such as power spikes, sudden current changes, and disconnected loads with a confidence level of 0.99–1.00. The integration of IoT, artificial intelligence, and basic protection systems results in a reliable and responsive monitoring device. In the future, this system has the potential to be developed for automatic efficiency analysis and intelligent load control.

Enhancing Polyp Segmentation Using Attention U-Net with CLAHE

Ramdaniah Ramdaniah — 2026-01-29

Colorectal cancer remains one of the leading causes of death worldwide, where early detection of polyps through colonoscopy plays a vital role in prevention. This study aims to enhance polyp segmentation performance by integrating Attention U-Net with Contrast Limited Adaptive Histogram Equalization (CLAHE) as a preprocessing technique. The proposed method was evaluated using two benchmark datasets, CVC-ClinicDB as the primary dataset and Kvasir-SEG for cross-domain testing. The model was trained using a combination of Binary Cross-Entropy and Dice losses, with a 70–15–15 split for training, validation, and testing. Experimental results show that applying CLAHE improves segmentation accuracy, achieving Dice and IoU scores of 0.84 and 0.76 on CVC-ClinicDB, and 0.62 and 0.50 on Kvasir-SEG, respectively. Statistical analysis using the Wilcoxon signed-rank test confirmed a significant difference between the baseline and enhanced models. These findings demonstrate that the integration of CLAHE with Attention U-Net effectively improves boundary detection and robustness against illumination variations across datasets, contributing to more accurate and reliable computer-aided diagnosis in colorectal cancer screening.

Bayesian-Optimized Prophet for Tourism-Based Regional Government Revenue Forecasting

Muhammad Sofwan Adha — 2026-01-29

Accurate hotel tax revenue forecasting is critical for supporting proactive fiscal planning in tourism-dependent local governments . Hotel tax revenues in these regions exhibit high volatility influenced by seasonal tourism patterns, visitor preferences, economic conditions, and external shocks such as the COVID-19 pandemic . Traditional time series forecasting methods such as Autoregressive Integrated Moving Average (ARIMA) and Exponential Smoothing struggle to capture complex seasonal patterns and accommodate multiple external factors . Recent advances in time series forecasting—particularly Facebook's Prophet framework—offer automatic decomposition of trend, seasonality, and holiday effects, plus the ability to integrate external regressors . However, Prophet's performance is highly sensitive to hyperparameter configurations, and default settings often produce suboptimal results on volatile data . Bayesian Optimization has emerged as an efficient technique for hyperparameter tuning, achieving convergence with significantly fewer iterations compared to exhaustive grid search . This study develops and validates a Bayesian-Optimized Prophet Framework for forecasting monthly hotel tax revenue in Kabupaten Tana Toraja, a cultural tourism destination in Indonesia, over 60 months (January 2020–December 2024) encompassing normal conditions, pandemic disruption, and recovery phases. The optimized model achieved Mean Absolute Percentage Error (MAPE) of 9.59% compared to baseline Prophet's 33.72%—a 71.55% improvement in forecasting accuracy. Mean Absolute Error (MAE) reduced from Rp 11.76 million to Rp 3.34 million per month. Robustness testing during COVID-19 pandemic demonstrated model stability with MAPE ≤15% despite >60% revenue decline. The framework provides 24-month forecasts (2025–2026) with 95% confidence intervals and decision-support capability with lead-time advantage of 3–6 months for early revenue shortfall detection. This research contributes a reproducible, efficient methodology for hyperparameter tuning in time series forecasting within fiscal planning domain, applicable to other tourism-dependent regions and tax categories.

Energy Efficient IoT-Based Forest Fire Detection Using LoRaWAN and AI

Muhammad Syafaat — 2026-01-29

Forest fires remain a global problem that has a major impact on the economy and health. Indonesia suffered losses of up to Rp. 72.95 trillion due to forest fires in 2019. Internet of Things (IoT) technology can be used for early detection of forest fires, but is constrained by limited network infrastructure and high energy consumption. This study aims to design a smart mitigation device and application for early detection of forest fires using LoRaWAN technology, which does not require an internet connection from the node to the gateway. In addition, an Artificial Intelligence method with adaptive sampling is applied, namely adaptive sampling threshold modeling and reinforcement Q-learning on the gateway to optimize energy use. The method used is Research and Development (R&D), with testing of the effectiveness of the design and descriptive statistical analysis to compare the energy efficiency between LoRaWAN devices with AI and conventional smart mitigation devices. The results of the study show that LoRa-based mitigation devices can cover the entire Jompie Botanical Garden area with a transmission distance of up to 3 kilometers and are 105% more energy efficient than conventional mitigation devices.

Daily Electricity Load Forecasting in Ternate City Using ELM

Andi Muhammad Ilyas — 2026-01-29

The continuously increasing growth of electricity demand necessitates accurate and systematic planning of electric power systems to ensure power flow quality and system reliability. Ternate City, as one of the major activity centers in North Maluku Province, has experienced a substantial rise in electricity consumption, thereby requiring an effective and reliable load forecasting approach. This study aims to predict the daily electricity load in Ternate City using the Extreme Learning Machine (ELM) method. The analysis is conducted using historical electricity load data, which are processed through data preprocessing stages, dataset partitioning into training and testing sets, and ELM-based modeling. The performance of the proposed model is evaluated using the Mean Absolute Percentage Error (MAPE). The results indicate that the MAPE values for the training dataset range from 5.84% to 13.63%, corresponding to very good to good performance categories. Meanwhile, the testing dataset yields MAPE values ranging from 13.45% to 33.09%, which fall within the good to sufficient performance categories. Furthermore, the prediction results are able to accurately capture daily electricity load fluctuation patterns from Monday to Sunday, including peak load periods. Based on these findings, the ELM method demonstrates strong potential as a reliable approach to support electric power system planning and to enhance the quality and reliability of electricity supply in Ternate City.

Augmented Reality and Virtual Reality in English Learning: Bibliometric Analysis of Research Trends, Citation Patterns, and Future Directions

Tamra Tamra — 2026-01-29

This study conducts a comprehensive bibliometric analysis to map the development of research on Augmented Reality (AR) and Virtual Reality (VR) in English language learning (ELL) from 2010 to 2025. Using 386 Scopus-indexed documents, the analysis examines publication growth, citation performance, influential authors and countries, core sources, and the thematic evolution of immersive learning research. The findings show a sharp increase in scientific production after 2020, reflecting the global rise of digital and immersive technologies in education. China, Korea, and Malaysia emerge as dominant contributors, demonstrating Asia’s leading role in AR/VR-driven language innovation. Citation trends reveal the coexistence of foundational highly cited works and rapidly influential recent publications. Source impact analysis confirms the interdisciplinary character of the field, spanning educational technology, linguistics, psychology, and computer science. Trend-topic analysis indicates a shift from general pedagogical themes toward AI-enhanced AR applications, deep learning, virtual reality environments, and interactive vocabulary learning systems. Despite significant growth, gaps remain in long-term studies, cross-country collaboration, and research on advanced language competencies. Overall, the study provides a data-driven understanding of how AR and VR have evolved as transformative tools for English language learning and offers strategic insights for guiding future research agendas in immersive educational technologies.

Performance Optimization of Image Cryptography for Copyright Protection on High-Resolution Images Using the Hill Cipher with Flexible Matrix Keys

Miftakhul Rohman — 2026-01-30

The increasing use of high-resolution digital images has raised serious concerns regarding copyright protection and unauthorized distribution. Image cryptography is one of the effective approaches to safeguard visual data by transforming images into unintelligible forms. The Hill Cipher algorithm, which is based on matrix operations, has potential for image encryption; however, its application to high-resolution images often suffers from high computational cost. This study proposes a performance optimization of image cryptography for copyright protection by exploiting the flexibility of matrix key sizes in the Hill Cipher algorithm. The optimization focuses on improving computational efficiency without modifying the fundamental cryptographic mechanism. Experiments were conducted on high-resolution images using different matrix key sizes (2×2, 3×3, and 4×4). Performance was evaluated in terms of encryption and decryption time, while security robustness was assessed using Entropy, Number of Pixel Change Rate (NPCR), and Unified Average Changing Intensity (UACI). The experimental results demonstrate that increasing the matrix key size significantly reduces the total computation time, achieving up to nearly 50% performance improvement, while maintaining high security levels. The encrypted images exhibit entropy values close to the ideal level, NPCR values above 99%, and stable UACI values, indicating strong randomness and diffusion properties. These findings confirm that the proposed optimization improves computational performance without compromising cryptographic security. Therefore, the optimized Hill Cipher remains effective and suitable for copyright protection of high-resolution images.