TIN: Terapan Informatika Nusantara

Sistem Informasi Arsip Berkas Perkara dengan Replikasi Basis Data dan Deteksi Anomali VAE

Sun, 23 Aug 2026 00:00:00 +0700

The management of case-file archives in prosecutorial institutions requires accurate tracking, controlled access, and reliable protection of archival records throughout the pre-prosecution, prosecution, and execution stages. Existing archival procedures rely on retention schedules and access-control mechanisms, but the management and protection of archival information require digital support to improve traceability, data integrity, and resilience against unauthorized activities or data loss. This study develops a case-file archive information system that integrates digital archival metadata with security and data-resilience mechanisms. The proposed system combines two-factor authentication (2FA) using TOTP, Role-Based Access Control (RBAC), SHA-256 cryptographic hashing, immutable audit logs with hash chaining, database replication using a Primary-Replica architecture with Write-Ahead Logging (WAL) streaming, and anomaly detection using a Variational Autoencoder (VAE). The system was developed using an Evolutionary Prototype approach and implemented with Electron, ReactJS, Node.js, PostgreSQL, and PyTorch. Evaluation was conducted in an intranet environment with approximately 100 active users through User Acceptance Testing, API testing, integrity testing, database replication testing, and VAE performance evaluation using Precision, Recall, F1-Score, and AUC-ROC. The results showed that all 13 functional scenarios and 16 API endpoints operated as expected, integrity mechanisms successfully detected all data manipulations in the test data, and database replication was synchronized automatically. The VAE achieved a Recall of 1.0000, F1-Score of 0.8197, and AUC-ROC of 1.0000, while its Precision of 0.6944 indicates that further threshold optimization and model refinement are required to reduce false-positive detections. The system therefore serves as digital support for existing archival procedures rather than replacing established archival practices in prosecutorial institutions.

Pengembangan Sistem Manajemen Surat Berbasis Web Menggunakan Design Thinking dan Agile-Scrum

Mon, 10 Aug 2026 00:00:00 +0700

Digital transformation in higher education has increased the demand for more effective, efficient, and integrated correspondence administration systems. However, the Faculty of Communication and Information at Universitas Garut continues to experience problems in managing incoming and outgoing correspondence because letter numbering, document distribution, disposition, and archiving processes are still performed manually and are not yet integrated. These conditions result in administrative delays, an increased risk of recording errors and data duplication, difficulties in retrieving archived documents, and reduced efficiency of administrative services. This study aims to develop a web-based Mail Management System by applying the Design Thinking approach to User Interface (UI) and User Experience (UX) design while adopting the Agile-Scrum framework as an iterative and adaptive software development methodology. The research stages consisted of user needs identification, prototype design, system development, feature implementation, and usability evaluation using the System Usability Scale (SUS). The developed system provides integrated features, including automated letter numbering, incoming and outgoing mail management, user and access management, letter templates, period and organizational unit management, letter disposition, and digital archiving. The usability evaluation involving six respondents produced an average SUS score of 70, which falls into the Acceptable category with a Grade C, indicating that the system has good usability and is acceptable to users. The findings of this study indicate that the integration of Design Thinking and Agile-Scrum is an effective approach for developing a web-based Mail Management System that is aligned with user needs.

Sistem Pemantauan dan Pengawasan Keamanan Siswa Berbasis Wearable IoT di Lingkungan Sekolah Madrasah Ibtidaiyah

Aqmal Fahat Syafiq Putra, Eka Arriyanti, Kusnandar Kusnandar — Mon, 10 Aug 2026 00:00:00 +0700

Student safety within school environments remains a critical concern, as manual supervision is constrained in its capacity to monitor students’ whereabouts and conditions rapidly and continuously. This study contributes by providing a technology-based medium that integrates IoT-enabled wearable devices, real-time data communication, and a centralized monitoring dashboard to strengthen school surveillance systems, while simultaneously addressing a gap in the literature concerning the implementation of wearable IoT in the context of Indonesian madrasah ibtidaiyah (elementary Islamic schools). The study aims to design and develop a wearable IoT-based student safety monitoring and supervision system capable of tracking students’ locations and basic physiological conditions in real time and transmitting emergency notifications to school authorities. The research was conducted at MI Al Azhar, Samarinda, East Kalimantan, from 20 July to 25 August 2026, involving six respondents: one principal, one teacher, one student affairs coordinator, one school IT technician, and two sixth-grade students. A Research and Development (R&D) approach employing a prototype model was utilized, encompassing stages of needs analysis, design, implementation, testing, and evaluation. Findings indicate that the system is capable of monitoring students’ conditions and whereabouts in real time, detecting potential risks via wearable devices, displaying movement histories and student status on a dashboard, and delivering emergency notifications more rapidly. The system also demonstrably enhances the effectiveness of student safety supervision in schools, as information can be monitored centrally and responses to incidents become more targeted. Based on testing outcomes and respondent feedback, the system is deemed feasible for use as an innovative solution to support student safety within school environments. The study’s contributions include the provision of an adaptable wearable IoT implementation framework for other schools, the strengthening of the empirical basis regarding the effectiveness of student monitoring systems at the elementary level, and the formulation of technical design recommendations and operational policies to support scalability and secure data governance.

Perancangan Prototype User Interface dan User Experience Marketplace Tanaman Online untuk UMKMM Menggunakan Metode Design Thinking

Dika Sartika, Fadlina Fadlina, Dwi Pertiwi Anggraini — Sun, 23 Aug 2026 00:00:00 +0700

The advancement of information technology has encouraged Micro, Small, and Medium Enterprises (MSMEs) to utilize digital media to improve product marketing and sales. Kebun Berkah Abu Abdillah still conducts its sales process through social media, resulting in unsystematic product information and manual transaction processes. This study aims to design a User Interface (UI) and User Experience (UX) prototype for an online plant marketplace using the Design Thinking method with the assistance of the Figma application. Data were collected through observations, interviews, and literature studies to identify user requirements. The Design Thinking process consisted of five stages: Empathize, Define, Ideate, Prototype, and Testing. The resulting prototype was evaluated using the System Usability Scale (SUS) involving 20 respondents. The evaluation produced an average SUS score of 71.5, which is above the standard benchmark score of 68, indicating that the prototype falls into the Good, Grade C+, and Acceptable categories. The results show that the prototype has a good level of ease of use, is easy for users to understand, and is suitable to serve as a reference in developing a marketplace system to support the digital sales of plants at UMKM Kebun Berkah Abu Abdillah. This study contributes by providing a design of an online plant marketplace UI/UX prototype that is created based on user needs using the Design Thinking method and has been validated through System Usability Scale (SUS) testing. The resulting prototype can be used as a reference in developing a marketplace system for SMEs engaged in plant sales.

Analitik Prediktif Terintegrasi: Regresi Linear Berganda untuk Keputusan Pembelian dan MA-3 untuk Peramalan Pendapatan

Novita Sambo Layuk, Asrul Syam, Samsu Alam, Santi Santi — Sun, 23 Aug 2026 00:00:00 +0700

This study integrates purchase-decision analysis and short-term revenue forecasting to support data-driven management at the Thrift Store. The population consisted of 1,536 customers; the minimum sample size calculated using Slovin’s formula at a 10% margin of error was 94, while 100 respondents were analyzed. Questionnaire data were examined using Multiple Linear Regression, and monthly revenue data from July to December 2025 were analyzed using a 3-Month Moving Average (MA-3). All questionnaire items were valid (r_hitung=0.7590–0.8569 > r_table=0.1966) and reliable (Cronbach’s Alpha=0.855–0.890). The residuals were normally distributed (Asymp. Sig.=0.605), multicollinearity was not detected (VIF=2.0460–2.5422), and the Durbin-Watson statistic was 2.002; however, the Glejser test indicated heteroscedasticity for price (p=0.001), so conventional inference for this predictor should be interpreted cautiously. OLS estimates showed positive coefficients for product quality (β=0.1998; p=0.0146), price (β =0.2255; p=0.0044), and service (β =0.4740; p<0.001), while the overall model was significant (F=61.084; p<0.001; R²=0.656). Service had the strongest bivariate correlation with purchase decisions (r=0.742). MA-3 produced a January 2026 revenue forecast of IDR 41,464,733, with a historical Mean Absolute Deviation (MAD) of IDR 18,522,744, indicating substantial forecast uncertainty. The contribution of this study lies in integrating cross-sectional analysis using Multiple Linear Regression and time-series analysis using MA-3 within a unified data-driven decision-making framework that combines perspectives on consumer behavior and revenue dynamics. Managerially, service improvement and transparent product information should be prioritized, while MA-3 is more appropriate as an early monitoring tool than as a high-precision forecasting model.

Pengaruh Literasi Keuangan dan Gaya Hidup Terhadap Perilaku Konsumtif Mahasiswa

Sun, 23 Aug 2026 00:00:00 +0700

Students’ consumerist behavior remains a problem influenced by low financial literacy and a lifestyle that tends to follow trends. This study aims to determine the influence of financial literacy and lifestyle on the consumerist behavior of students at Alauddin State Islamic University in Makassar. This study employs a quantitative approach using a causal-associative research design. Data were collected through a questionnaire distributed via Google Forms to 32 students selected using purposive sampling. The research instrument used a Likert scale and was analyzed using multiple linear regression with the assistance of the SPSS application. The results indicate that, simultaneously, financial literacy and lifestyle significantly influence students’ consumption behavior, with a significance value of 0.002 < 0.05 and a calculated F-value of 7.647. The coefficient of determination of 0.345 indicates that financial literacy and lifestyle can explain 34.5% of consumptive behavior, while the remainder is influenced by other factors outside the scope of this study. Partially, financial literacy has a significant effect on consumptive behavior, whereas lifestyle does not have a significant effect.

Model Konseptual Penguatan Instrumen Penilaian Kematangan Keamanan Siber (IKAS) menggunakan Pendekatan Sosioteknis melalui Pemetaan Keselarasan terhadap NIST CSF 2.0 dan ISO/IEC 27001:2022

Muhammad Arif Ali Wasi, Agung Budi Susanto, Winarni Winarni — Sun, 23 Aug 2026 00:00:00 +0700

The Cybersecurity Maturity Assessment Instrument (IKAS) underpins the protection roadmap of Vital Information Infrastructure in Indonesia, so any weakness in its design propagates directly into national policy. Its alignment with international frameworks and the balance of its sociotechnical composition, however, have never been examined academically. This study maps the alignment of IKAS version 1.2.1 against NIST CSF 2.0 and ISO/IEC 27001:2022 bidirectionally, diagnoses the gaps through Leavitt’s Diamond, and designs a conceptual strengthening model named IKAS-ST. A descriptive qualitative content analysis was applied to 181 IKAS items, 106 NIST CSF 2.0 subcategories, and 93 ISO/IEC 27001:2022 Annex A controls. The instrument proves substantially aligned, with an average forward coverage of 81.2% (Largely Achieved), yet 12 white spots remain, concentrated in the governance and recovery functions. The decisive finding is that Task is the only under-represented scope, with an extreme deficit in the Detection domain where items measure the ownership of detection technology rather than the procedures that operate it. The merit of IKAS-ST lies not in adding items but in three testable properties: every added item is traceable to an identified white spot, so no addition is speculative; the restructuring addresses the root cause, namely the absence of an explicit procedural scope; and all 181 original items are retained, preserving the comparability of historical assessment results. The model is conceptual and awaits expert validation and field testing.

Analisis Dampak Route Summarization Terhadap Performa OSPF Multi-Area Menggunakan GNS3 MikroTik

Muhammad Ilfan Asykuri, Zainal Arifin, Mochammad Faid — Sun, 23 Aug 2026 00:00:00 +0700

Open Shortest Path First (OSPF) multi-area networks at medium to large scales face the problem of Link State Database (LSDB) bloat caused by uncontrolled Link State Advertisement (LSA) propagation, particularly when each subnet is individually advertised by the Area Border Router (ABR) across all network areas. This condition increases memory consumption, the computational load of the Shortest Path First (SPF) algorithm, and routing convergence delays. This study aims to quantitatively analyze the impact of applying route summarization on the ABR regarding the number of LSAs per type, LSDB size, routing table entries, convergence time, and CPU and memory utilization of MikroTik RouterOS CHR devices in an OSPF multi-area environment. The method employed is quantitative experimental with two controlled scenarios: without route summarization as a baseline and with route summarization on the ABR as the treatment condition. The topology was built on the GNS3 simulation platform using 12 MikroTik CHR routers distributed across three areas with 24 active subnets and 1,000 dummy prefixes. Results show that route summarization reduced Type-3 LSAs by 99.71%, decreased LSDB size by 99.32%, reduced routing table entries by 99.22%, lowered peak ABR CPU usage from 11.33% to 8.33%, and achieved perfect fault isolation on the Area 0 control plane with an average convergence time of 0.2 ms compared to 1,023 ms in the baseline condition. This study contributes practical guidelines for network administrators to optimize router memory utilization and accelerate convergence in multi-area network architectures.

Deteksi Defect Sepatu Berbasis Fine-Tuned CNN Menggunakan Arsitektur DenseNet untuk Mendukung Proses Quality Control

Sun, 23 Aug 2026 00:00:00 +0700

Indonesia is a leading global exporter of footwear. Despite successfully exports million pair of shoes, shoes industry in Indonesia faces significant challenges due to conventional quality control (QC) limitations. This study proposes a deep learning-based defect detection model to enhance the efficiency and accuracy of the QC process. This study uses primary datasets collected from Laboratory of PIDI 4.0 Politeknik ATK Yogyakarta and employed in four experimental scenarios. The dataset consists of 2,828 images comprising of two classes (good class and reject class). This study utilizes fine-tuned CNN using DenseNet architecture for distinguishing two classes. Before employing the fine-tuned CNN method, we separate the data into three components, which are training, validation and testing data, with proportion of 80:10:10. The results demonstrate that the integration of data augmentation and the Adam optimizer yielded the highest performance, achieving accuracy, precision, recall and F1-score of 0.9546, 0.9551, 0.9546, 0.9545, respectively. These findings suggest that implementing automated deep learning models has potential to reduce rejection rates by modernizing traditional inspection methods, which can strengthen the competitiveness of Indonesia’s footwear industry in the global market.

DenseNet121 sebagai Feature Extractor pada Denoising Autoencoder untuk Deteksi Anomali Unsupervised Citra X-Ray Dada

Sun, 23 Aug 2026 00:00:00 +0700

Anomaly detection in chest X-ray images remains a challenge in medical imaging, as supervised learning approaches require large amounts of labeled data that are difficult and costly to annotate. This study proposes an unsupervised learning anomaly detection system that integrates a pretrained DenseNet121 as a feature extractor with a Denoising Autoencoder (DAE), so that training requires only normal images without anomaly annotations. The model was trained on normal images and tested on COVID-19 and pneumonia images to evaluate its anomaly detection capability based on reconstruction error relative to an optimal threshold. Evaluation was conducted on the Covid19-Pneumonia-Normal Chest X-Ray Images dataset comprising 5,228 images, comparing the performance of DenseNet121 and ResNet50 as feature extractors across three latent dimension configurations. The DenseNet121 configuration with a latent dimension of 128 achieved the highest overall performance on most metrics, namely 91% accuracy, 90.75% sensitivity, 72% Macro F1-Score, and a validation loss (MSE) of 0.0952 on 3,608 test images, although its AUC (0.9526) and specificity (87.36%) were not consistently the highest among all tested configurations. These results demonstrate that using DenseNet121 as a feature extractor improves the DAE's ability to distinguish normal from anomalous lung images, suggesting its potential as an efficient preliminary screening approach under conditions of limited labeled data.

Data-Driven Business Process Improvement untuk Mendukung Optimasi Proses Bisnis pada UMKM Menggunakan Process Mining

Novia Andini, Haerul Haerul — Sun, 23 Aug 2026 00:00:00 +0700

In this digital era, technology has become an essential requirement for Micro, Small, and Medium Enterprises (MSMEs) to support business competitiveness. However, limitations in information systems often result in data being stored in non-standardized spreadsheet formats, causing the data to remain "dirty" and unprepared for process analysis. This poses a significant technical challenge in transforming raw data into valid event logs, particularly regarding timestamp synchronization, accurate Case ID mapping from manual records, and process flow adjustments. Many MSMEs still lack such standardization and rely on conventional approaches to Business Process Improvement (BPI), where decision-making is based on subjective and less accurate interviews or observations. Therefore, this study aims to develop a data-driven BPI approach by incorporating process mining to assist in the optimization of MSME business processes. The methodology of this research includes collecting MSME business process activity data in the form of digital event logs using spreadsheets, analyzing data and processes through mining techniques, identifying bottlenecks and inefficiencies, and designing business process improvements. The data used consists of operational activity records from a traditional herbal medicine (Jamu) MSME in Cimahi, which has been simply digitized using spreadsheets. The novelty of this research lies in the application of data-driven process mining for BPI using MSME spreadsheet event logs. The results of this study, processed from 6,225 events and 1,027 cases, reveal two process variants: a normal variant (93.87%) with an average duration of 5 hours 6 minutes, and a variant involving rework (6.13%) with an average duration of 6 hours 41 minutes. Performance analysis identified bottlenecks in the transition from Production to Packaging (1 hour 55 minutes) and from Shipping to Completed Order (1 hour 53 minutes). A new process model is proposed to eliminate "Rework," which is projected to reduce the overall average duration to 4 hours 42 minutes. The contribution of this research is the development of a data-driven BPI approach based on process mining that can be applied to MSMEs with limited information systems through the utilization of spreadsheet-based event logs. This approach enables the identification of process variations and bottlenecks based on actual operational data and provides a measurable basis for designing improved business processes. The results of the study indicate that this approach can be used to support business process optimization in MSMEs without relying on complex information systems.

Komparasi Efektivitas PCC dan ECMP pada Arsitektur Multi-ISP Heterogen Ditinjau dari Parameter Quality of Services (QoS)

Reza Dwi Cahya Kurniawan, Riska Wibowo — Sun, 23 Aug 2026 00:00:00 +0700

Fast and stable internet access is essential to support academic and administrative activities at Universitas Muhammadiyah Tegal. To overcome bandwidth fluctuations and ensure optimal connectivity, implementing a multi-ISP load balancing system is required. This study aims to compare the performance of Per Connection Classifier (PCC) and Equal-Cost Multi-Path (ECMP) load balancing algorithms based on TIPHON Quality of Service (QoS) standards at Universitas Muhammadiyah Tegal. The methodology utilizes a quantitative experimental approach on a MikroTik router with a dual-WAN configuration (2 ISPs × 100 Mbps). Testing was conducted over 7 days during peak hours (08:00–16:00 WIB) using iPerf3 for traffic injection and Wireshark for packet analysis. Quantitative results reveal that the PCC algorithm achieves superior performance across all QoS parameters, recording a throughput of 194.8 Mbps (97.4% efficiency), a delay of 28.5 ms, a jitter of 3.4 ms, and a packet loss of 0.3%. In contrast, ECMP produced a jitter of 11.6 ms and a packet loss of 2.8%, triggering session drops on critical services. PCC’s advantage is driven by its session locking mechanism via precise mangle rules, effectively mitigating bufferbloat and hash collisions. Based on a comprehensive evaluation of throughput, delay, jitter, and packet loss, the PCC algorithm is objectively superior and strongly recommended for implementation at Universitas Muhammadiyah Tegal.

Integrasi Fuzzy Mamdani dan Certainty Factor pada Sistem Pakar Prediksi Penyakit Jantung

Naza Riski Romah Doni, Hadi Zakaria — Sun, 23 Aug 2026 00:00:00 +0700

Heart disease is a leading cause of death globally; therefore, a rapid and structured initial screening process is essential to minimize the risk of delayed treatment. The PT. XYZ clinic faces challenges in conducting initial heart disease screenings due to limited diagnostic equipment and an identification process that relies heavily on the subjective assessment of medical personnel. This study aims to implement a web-based expert system that integrates the Fuzzy Mamdani method and the Certainty Factor method to assist in the initial screening of heart disease. The Fuzzy Mamdani method processes symptom values into membership degrees and generates alpha-predicate values through fuzzification and inference processes, while the Certainty Factor method calculates the confidence level based on alpha-predicate values and expert-assigned confidence weights for each rule. The system's knowledge base comprises 14 symptoms, three types of heart disease, and 190 rules derived through a knowledge acquisition process with cardiologists. The contribution of this study lies in the implementation of Fuzzy Mamdani and Certainty Factor integration, in which the alpha-predicate value resulting from fuzzy inference is used as the rule activation level and combined with the expert CF to determine the system confidence level for initial heart disease screening. Testing was conducted using 30 case scenarios and evaluated via a confusion matrix. The results demonstrate that all scenarios were successfully classified according to the reference labels, achieving a 100% match rate with the established knowledge base. These findings indicate that the system consistently executes inference and confidence level calculations as a tool for initial heart disease screening, although it is not intended to replace a physician's diagnosis or clinical decision-making.

Perancangan UI/UX Aplikasi Pendaftaran Siswa RA Berbasis Web Menggunakan Design Thinking dan Evaluasi SUS

Amelia Khairani; Fince Tinus Waruwu, Melda Panjaitan — Sun, 23 Aug 2026 00:00:00 +0700

The student admission process at RA Tahfiz Qur'an Darussalam is still conducted manually, resulting in inefficient information dissemination, form completion, document submission, and data management, while also increasing the risk of recording errors. This study aims to design a User Interface (UI) and User Experience (UX) for a web-based student admission application that meets user needs. The research employed the Design Thinking method, consisting of the Empathize, Define, Ideate, Prototype, and Test stages. The Empathize stage was conducted through observations and interviews to identify user needs, followed by the Define stage to formulate design solutions. The proposed design was then visualized into wireframes and high-fidelity prototypes using Figma during the Prototype stage. The usability evaluation was carried out using the System Usability Scale (SUS) involving 30 respondents, consisting of parents, guardians of prospective students, and a school administrator. The evaluation results produced an average SUS score of 88.92, which falls into the Acceptable category, achieves Grade Scale B, and is classified as Excellent on the Adjective Rating scale. These findings indicate that the proposed UI/UX design has a high level of usability, is easy to understand, and effectively supports users in completing the web-based student admission process.

Early Warning System (EWS) Kualitas Air Hidroponik Melon Berbasis IoT dan Sensor pH - TDS - Suhu

Lilis Kurlina, Amalia Herlina, Zainal Arifin — Sun, 23 Aug 2026 00:00:00 +0700

Dependence on handheld instruments in managing hydroponic melon cultivation at Tanjung Farm is a major constraint, as it is unable to detect changes in water conditions rapidly and continuously, thereby increasing the risk of crop failure due to delayed intervention. This study aims to develop an IoT (Internet of Things)-based nutrient solution quality monitoring system capable of measuring four water parameters in real-time, namely pH (potential of Hydrogen), TDS (Total Dissolved Solids), turbidity, and temperature, equipped with an automatic Early Warning System (EWS). The study adopted a Mixed Methods design with a prototype development approach conducted at Tanjung Farm during 2025–2026, involving 2 respondents consisting of the farm supervisor and farm manager as system end-users. The system was constructed using an ESP32 microcontroller as the central processing unit, four water quality sensor modules, a MySQL database for data archiving, and a web interface developed using the Laravel framework. Performance evaluation revealed that the pH sensor and DS18B20 temperature sensor achieved high precision with a 0% error rate compared to handheld measuring instruments, while the TDS sensor produced a 26.7% error attributed to the absence of a hardware temperature compensation module. Average data transmission latency from the microcontroller to the server was recorded at 5 seconds with EWS notification updates every 30 seconds. The resulting system successfully shifted conventional monitoring patterns toward an integrated digital solution operable remotely to maintain the productivity stability of hydroponic melon cultivation.

IoT-Based Voice Communication Control System for ATC Radar Simulator Support

Ayub Wimatra, Sunardi Sunardi, Julfansyah Margolang, Catra Indra Cahyadi — Sun, 23 Aug 2026 00:00:00 +0700

The Voice Communication Control System (VCCS) used in the Air Traffic Control (ATC) Radar Simulator plays an essential role in supporting communication between ATC trainees and pseudo pilots during simulation-based training. However, the existing proprietary VCCS faces obsolescence issues, limited component availability, and high maintenance costs, creating challenges for long-term operational sustainability. This study proposes an Internet of Things (IoT)-based VCCS prototype as an alternative communication control solution using embedded wireless control and Android-based human-machine interfaces. The system was developed using a prototype-based approach consisting of requirements identification, conceptual architecture design, implementation, and iterative testing. Functional testing was conducted to evaluate core communication control features, followed by reliability and performance evaluation through 100 operational testing cycles. The results show that the developed prototype successfully fulfilled all primary functional requirements, including wireless connectivity, channel switching, push-to-talk control, digital volume adjustment, and voice communication. Reliability testing demonstrated an operational reliability of 96%, while cost analysis showed a cost efficiency of 91.05% compared to conventional proprietary VCCS systems. These findings indicate that the proposed IoT-based VCCS provides a practical, reliable, and economically sustainable solution for communication control modernization in aviation training environments.

Implementasi Hybrid Feature Fusion menggunakan ResNet50 dan RGB-HSV untuk Klasifikasi Indikasi Tingkat Kepedasan Cabai Berdasarkan Karakteristik Visual

Febriyan Biopsa Minanda — Sun, 23 Aug 2026 00:00:00 +0700

Chili is a horticultural commodity with diverse visual characteristics and different levels of pungency, requiring an objective and consistent classification method. A challenge in chili classification is the similarity of visual characteristics across categories and the separate use of color features or deep learning features in previous studies, resulting in less comprehensive information representation. This study aims to implement hybrid feature fusion using ResNet50 and RGB-HSV color features to classify indications of chili pungency into Hot, Medium, and Mild categories based on visual characteristics. The novelty of this study lies in integrating deep features from ResNet50 with RGB-HSV features into a single feature representation, as ResNet50 can capture complex visual patterns, while RGB-HSV explicitly provides color information. This study employed a quantitative approach with an experimental methodology based on image processing and deep learning. The dataset consisted of 1,592 chili images, including 416 Hot, 514 Medium, and 662 Mild images. The data were divided at a 70:15:15 ratio into 1,114 training, 239 validation, and 239 testing images. Preprocessing included background removal using rembg, resizing images to 224 × 224 pixels, and extracting mean and standard deviation features from RGB and HSV channels. The model achieved 96.23% accuracy, 96.10% precision, 96.08% recall, and a 96.08% F1-score, outperforming the original image model. These findings indicate that integrating deep and color features improves chili classification performance.

Analisis Data Transaksi Konsumen Menggunakan Algoritma Apriori untuk Strategi Promosi pada Grosir Sembako

Vera Oktaviani, Joko Priambodo — Sun, 23 Aug 2026 00:00:00 +0700

Wholesale grocery businesses generate an increasing amount of consumer transaction data, but such data has not been optimally utilized to identify purchasing patterns and relationships among products as a basis for developing promotional strategies. This condition means that promotional decisions may still be made based on general observations of frequently sold products, making it difficult to systematically identify opportunities from consumer purchasing patterns. This study aims to analyze the relationships among products in consumer transaction data using the Apriori algorithm and to generate promotional strategy recommendations in the form of bundling and cross-selling. The data consist of 350 consumer transactions involving nine product categories. The 350 transactions were used because they represent the entire transaction data available during the research observation period, thereby reflecting the transaction conditions of the research object during that period. Data processing was conducted using the Knowledge Discovery in Databases (KDD) stages, namely selection, preprocessing, transformation, data mining, and evaluation. The Apriori algorithm was applied with a minimum support of 30% and a minimum confidence of 70%. The analysis produced 9 frequent 1-itemsets, 21 frequent 2-itemsets, 3 frequent 3-itemsets, and 22 association rules that met the minimum thresholds. The rule with the highest confidence was Oil and Flour → Eggs, with a support of 48.57% and a confidence of 92.39%. The validation results showed that the manual calculations and system results produced consistent values. Based on the obtained association rules, promotional strategy recommendations can be directed toward bundling and cross-selling according to the confidence level of each rule. This study demonstrates that the Apriori algorithm can be used to identify purchasing patterns from wholesale transaction data and generate information that can support the development of data-driven promotional strategies. However, the results are limited to 350 transactions and do not directly measure the impact of implementing the promotional strategies on sales.

Perancangan Rekam Medik Elektronik Pasien Jiwa dan Evaluasi Usability dengan System Usability Scale (SUS)

Galank Rasta Fariant, Wahyu Teja Kusuma, M. Syauqi Haris — Sun, 23 Aug 2026 00:00:00 +0700

The development of information technology drives the transformation of medical record management into Electronic Medical Records (EMR), including in the rehabilitation services for patients with mental disorders, which require continuous recording of psychological conditions, therapy progress, and service history. At the Yayasan Ketenangan Jiwa dan Hati (YKJH), the recording process is still done manually, which has the potential to cause recording errors, delays in information access, and difficulties in managing patient data. This research aims to design an RME interface that meets the needs of users in mental health patient rehabilitation services and to evaluate the usability level of the prototype using the System Usability Scale (SUS). The research applies the Waterfall method, which includes requirements analysis, system design, prototype implementation, and usability evaluation. Data were collected thru observation, interviews, and literature review. The resulting prototype provides features such as login, dashboard, patient registration, patient data management, diagnosis, SOAP notes, prescriptions, medical history, reports, and user management. Usability evaluation was conducted on five respondents consisting of two administrative staff, two medical personnel, and one foundation manager thru scenarios using the main features of the system, followed by filling out the SUS questionnaire. The evaluation results obtained an average score of 84.0, which falls into the Excellent and Acceptable categories. The contribution of this research is to produce an RME design tailored to the needs of administration and rehabilitation services for patients with mental disorders, as well as to provide a usability evaluation as a basis for further system development. The research results indicate that the prototype has a good level of usability in the initial evaluation stage, although further testing with a larger number of users and in a direct operational environment is still needed.

Pengaruh Merger, Budaya Organisasi, dan Gaya Kepemimpinan Terhadap Komitmen Pegawai

CH Bagus Anggara Jita, Ike Kusdyah Rachmawati, Lussia Mariesti Andriany — Mon, 31 Aug 2026 00:00:00 +0700

The consolidation of Indonesian rural banks (Bank Perekonomian Rakyat; BPR) can strengthen institutional capacity but may also create uncertainty among employees before the merger becomes legally effective. This study examined the effects of merger, organizational culture, and leadership style on employee commitment at PT BPR Chandra Muktiartha. A quantitative cross-sectional survey design was employed. Likert-scale questionnaires were distributed to 138 employees; 64 were returned, of which 58 were complete and eligible for analysis. Each construct was measured using 12 items. Data analysis included descriptive statistics, validity and reliability testing, classical assumption tests, and multiple linear regression. Employee commitment was categorized as high (M = 4.149, SD = 0.452). The regression model was statistically significant, F(3, 54) = 8.242, p < .001, with an R² of .314. At the predictor level, organizational culture had a positive and significant effect on employee commitment (B = 0.474, p = .033). In contrast, neither merger (B = -0.072, p = .559) nor leadership style (B = 0.218, p = .124) had a significant effect. These findings indicate that, during the pre-merger stage, employees' day-to-day experience of organizational culture plays a more immediate role in shaping their commitment than structural changes that have yet to be implemented. Management should prioritize the internalization of organizational values, open communication, cross-unit collaboration, fairness, and consistent explanations regarding each stage of the merger process. The generalizability of the findings is limited by the cross-sectional design, non-probability sampling, and single-organization setting. The study contributes phase-specific evidence from a rural bank by testing the three predictors in one model and showing that organizational culture is the only significant partial predictor before the structural change is implemented.

Analisis Sentimen Terhadap Ulasan Google Review Honda Malang Sekawan Motor Menggunakan IndoRoBERTa

Fera Dhea Sagita, M. Syauqi, Risqy Siwi Pradini — Mon, 31 Aug 2026 00:00:00 +0700

Customer reviews on Google Review contain important information about service quality, but the large and unstructured amount of data makes manual analysis difficult. This research aims to analyze the sentiment of customer reviews on Google Review Honda Malang Sekawan Motor Singosari using the IndoRoBERTa model thru a fine-tuning process. The dataset was obtained thru web scraping, followed by preprocessing which included cleaning, case folding, normalization, tokenization, stopword removal, and stemming. Sentiment labeling is performed using the lexicon-based InSet method as the initial labeling (weak labeling), then the dataset is divided into training and testing data with an 80:20 ratio. The model is used to classify sentiment into three classes: positive, neutral, and negative. The test results show that the model achieved an accuracy of 0.94, precision of 0.92, recall of 0.94, and an F1-score of 0.92. The model shows good performance in classifying positive sentiment, but still has limitations in recognizing negative sentiment due to data distribution imbalance. The contribution of this research is to provide an evaluation of the application of fine-tuning IndoRoBERTa in the domain of Google Review reviews of Indonesian-language automotive dealers, which have different characteristics from previous studies that mostly used social media data, while also analyzing the influence of InSet-based weak labeling and class imbalance on classification performance. The research results can serve as a basis for utilizing sentiment analysis to help dealers understand customer perceptions and identify aspects of service that need to be maintained or improved. The evaluation of the research also shows that the classification results still represent the label patterns generated by InSet, so further research needs to use human annotation as ground truth to improve the validity of the results.

Implementasi Pengelompokan Citra Batik dengan Pendekatan Metode Graph-Based Multi-View Clustering

Budi Hartono, Veronica Lusiana — Mon, 31 Aug 2026 00:00:00 +0700

Batik is a cultural heritage of Indonesia characterized by a rich array of motifs, colors, geometric patterns, and complex textures. This visual complexity makes Batik images compelling subjects for analysis using image processing and machine learning approaches, particularly for unsupervised clustering. A primary challenge in this research is effectively representing Batik's visual characteristics so pattern similarities can be accurately identified. Three texture features contrast, entropy, and energy serve as distinct "views" of Batik characteristics. These feature values are obtained by extracting them using the Gray-Level Co-occurrence Matrix (GLCM) method. The study utilizes a Graph-based Multi-View Clustering approach to group Batik images based on inter-image similarity. The study constructs a combined consensus graph with 30 nodes and 219 edges by integrating contrast, energy, and entropy features. This graph exhibits a density of 0.503 and forms a single connected graph component. The density result indicates that 50.3% of all possible image pairs share a connection. The resulting graph forms a connected structure that incorporates information from all three views, yielding a more comprehensive clustering representation than any single view. The combined consensus graph establishes an interconnected structure that produces a representative clustering of Batik images.

Run-Level Comparison of Binary Swarm Optimizers for Gallstone Disease Feature Selection

Jabesh Nehemiah Wijaya — Mon, 31 Aug 2026 00:00:00 +0700

Gallstone disease prediction from non-imaging clinical variables can support early risk assessment, but redundant measurements may reduce model efficiency and complicate interpretation. This study compares five binary swarm optimizers for wrapper feature selection on the UCI Gallstone dataset: Golden Jackal Optimization (bGJO), Egret Swarm Optimization (bESOA), African Vultures Optimization (bAVOA), Tuna Swarm Optimization (bTSO), and FOX Optimization (bFOX). The main contribution of this study is a reproducible run-level statistical comparison of five binary swarm optimizers under an identical wrapper evaluation framework, providing evidence on their accuracy–parsimony and computational trade-offs for gallstone feature selection. The dataset contains 319 complete records, 38 predictors, and an approximately balanced binary outcome. Candidate subsets were evaluated with standardized five-nearest-neighbor classification and stratified 10-fold cross-validation. A weighted objective combined classification error (0.9) and selected-feature proportion (0.1). Each optimizer used 100 epochs, a population of 10, and 10 independent seeded runs. Run-level comparisons used Kruskal–Wallis tests followed by Holm-adjusted Mann–Whitney tests. bGJO achieved the highest mean accuracy (77.56%) with 13.0 features, whereas bAVOA obtained the best mean fitness (0.2270) with only 3.7 features. bTSO selected the fewest features (2.8) but showed greater variability, and bFOX produced the lowest accuracy (67.68%). Accuracy, fitness, feature count, runtime, and memory differed globally among algorithms (all p<0.001). The findings demonstrate a clear accuracy–parsimony trade-off: bGJO is preferable when predictive accuracy dominates, while bAVOA offers the strongest overall compromise under the specified objective. External validation and stability analysis of selected variables are required before clinical use.

Deteksi Anomali pada Citra X-ray Dada Menggunakan Variational Autoencoder dengan Skema Sequential Hyperparameter Optimization

Mon, 31 Aug 2026 00:00:00 +0700

The limited availability of labeled medical images remains a major challenge in developing reliable deep learning-based disease detection systems. Conventional classification approaches generally require a large amount of abnormal data, whereas medical image annotation is time-consuming, costly, and highly dependent on radiological expertise. This study proposes an unsupervised anomaly detection model for chest X-ray images using a Variational Autoencoder (VAE), in which only normal images are utilized during the training process. The experiments were conducted on the COVID-19-Pneumonia-Normal Chest X-ray Images dataset, consisting of 5,228 images categorized into normal, pneumonia, and COVID-19 classes. The proposed framework includes image preprocessing, baseline VAE construction, sequential hyperparameter optimization, Beta-VAE implementation, and model evaluation using Accuracy, Precision, Recall, F1-Score, and the Area Under the Receiver Operating Characteristic Curve (AUROC). Experimental results demonstrate that the optimized model outperformed the baseline model, achieving an Accuracy of 97.50%, Precision of 96.32%, Recall of 100%, F1-Score of 98.12%, and an AUROC of 0.9999. These findings indicate that hyperparameter optimization and appropriate β coefficient selection improve latent representation learning, leading to more effective discrimination between normal and abnormal chest X-ray images. Therefore, the proposed approach has the potential to serve as an artificial intelligence-based early screening tool for chest radiograph analysis, particularly in scenarios where labeled medical data are limited.

Pengembangan Sistem Informasi dan Monitoring Inventaris Berbasis Mobile untuk Meningkatkan Efektivitas Pengelolaan Aset

Refandi Asiking, Antonius P.G. Manginsela, Harson Kapoh — Mon, 31 Aug 2026 00:00:00 +0700

Advances in information technology are driving companies to implement digital systems for inventory management and asset monitoring. PT PLN Nusa Daya’s Sulawesi 1 Operational Unit in Tikala still relies on manual processes, which result in reporting delays, recording errors, and limitations in monitoring the condition and location of assets. This study aims to develop a mobile-based inventory information and monitoring system (SI-INVADA) using Flutter to improve the effectiveness of asset management. The method used is the Agile approach with a simplified Scrum framework, including the creation of a product backlog, the execution of sprints, and sprint reviews. Data collection techniques were conducted through observation, documentation, and literature review. System testing utilized Black Box Testing to ensure functional compliance and the System Usability Scale (SUS) to evaluate usability. The results showed that all system functions operated as required, with a 100% success rate in Black Box Testing and an average system response time of 1.37 seconds. The SUS evaluation yielded an average score of 86.7, which falls into the “Acceptable” category, with an “Excellent” rating and a grade of B. The system provides features for inventory management, monitoring of asset condition and location, maintenance history, and an integrated reporting dashboard. Thus, SI-INVADA has met functional and usability requirements and provided a solution to the challenges of manual asset management in terms of record-keeping, information retrieval, and report generation.

Pengaruh Kompetensi dan Budaya Organisasi terhadap Efektivitas Koordinasi Penanganan Perkara Koneksitas

Wisnu Muhammad Nuur Huda, Hery Erdi — Mon, 31 Aug 2026 00:00:00 +0700

This study analyses the effect of competence and organizational culture on the effectiveness of coordination in handling connexity cases at the Deputy Attorney General for Military Crimes (JAMPIDMIL), Attorney General's Office of the Republic of Indonesia. Handling connexity cases requires coordination among the Prosecution Service, the Military Oditurat and the Military Police, yet practice still shows recurring delays, differing perceptions of authority and non-uniform administrative formats. A quantitative causal-associative approach was applied at the central office of JAMPIDMIL. From a population of 137 employees, purposive sampling identified 64 eligible employees, and 46 complete responses were analysed. Data were collected through a closed questionnaire of 51 items on a five-point Likert scale and analysed using multiple linear regression with SPSS, preceded by validity, reliability and classical assumption testing. The results show that competence has a positive and significant partial effect on coordination effectiveness (t = 2.699; sig. = 0.010), while organizational culture is not yet partially significant (t = 1.976; sig. = 0.055). Simultaneously, both variables have a positive and significant effect (F = 23.670; sig. = 0.000) with a coefficient of determination of 0.524. The study concludes that competence is the dominant determinant, whereas organizational culture functions as a collective foundation, consistent with the Islamic values of itqan, ta'awun and syura. This study offers three contributions. Theoretically, it extends the application of the competence and organizational culture constructs from the domain of individual performance to that of inter-organizational coordination effectiveness. Methodologically, it operationalises Damhuri’s public sector coordination framework, previously conceptual in nature, into a tested eighteen-item scale. Practically, it provides an empirical basis for prioritising human resource development in work units that carry a cross-institutional coordination mandate.

Peran Gender dalam Pengaruh Key Audit Matters terhadap Kualitas dan Fee Audit di Indonesia

Tiara Rani Santoso, Hesiya May Ghaisani, Afif Musthafa, Ronia Tambunan — Mon, 31 Aug 2026 00:00:00 +0700

This study aims to examine the effect of Key Audit Matters (KAM) on audit quality and audit fees and to investigate the moderating role of auditor gender. The implementation of KAM as part of audit reporting reform is expected to enhance transparency and the informational value of the auditor’s report, thereby reducing information asymmetry between management and stakeholders. This study employs a quantitative approach using purposive sampling and obtains 264 firm-year observations from publicly listed companies during the observation period. Data are analyzed using Structural Equation Modeling–Partial Least Squares (SEM-PLS) to test both direct and moderating relationships among variables. The results indicate that KAM has a positive and significant effect on both audit quality and audit fees. These findings suggest that more extensive and informative KAM disclosures reflect greater audit effort and more rigorous risk assessment in significant audit areas, which ultimately improves audit quality. In addition, the increased complexity and intensity of audit procedures associated with KAM disclosures lead to higher audit fees as an economic consequence of enhanced audit effort. The moderation analysis further reveals that female auditors strengthen the positive relationship between KAM and audit fees, indicating that higher levels of prudence and risk sensitivity contribute to more intensive audit procedures and higher audit pricing. However, auditor gender does not moderate the relationship between KAM and audit quality. This study provides empirical contributions to the literature on audit reporting effectiveness and highlights the role of auditor characteristics in shaping audit outcomes and their economic implications.

Perbandingan Kinerja Algoritma SVM dan Random Forest dalam Klasifikasi Tuberkulosis di Puskesmas Sandar Angin

Rinda Desfourtheen, Lastri Widya Astuti, Muhammad Haviz Irfani — Mon, 31 Aug 2026 00:00:00 +0700

Tuberculosis (TBC) remains a major infectious disease and public health problem in Indonesia, requiring approaches that can support accurate diagnosis. Machine learning can assist TBC classification based on patients’ clinical characteristics, but different algorithms may produce different performance. This study aims to compare the performance of Support Vector Machine (SVM) and Random Forest and determine the best-performing algorithm for TBC classification using patient data from Sandar Angin Public Health Center, Pagar Alam. The study used primary data from 201 patients recorded from January 2023 to October 2025, consisting of 14 variables: number, gender, age, chest pain, productive cough, cough duration, shortness of breath, fatigue, weight loss, body temperature, cold sweats, decreased appetite, sputum, and diagnosis. Model evaluation was conducted using an 80:20 data-splitting scheme and 5-Fold Cross Validation. In the 80:20 split, SVM achieved 98% accuracy, 94% precision, 100% recall, and 97% F1-score, while Random Forest achieved 93% accuracy, 88% precision, 94% recall, and 91% F1-score. Based on 5-Fold Cross Validation, SVM achieved the highest accuracy of 90%, while Random Forest achieved 88%. Overall, SVM demonstrated superior classification performance and was selected as the best-performing algorithm for TBC classification using patient data from Sandar Angin Public Health Center, Pagar Alam.

Klasterisasi Indikator Data Kemiskinan Kabupaten/Kota di Indonesia Menggunakan K-Means dengan Normalisasi Z-Score dan Evaluasi DBI

Mon, 31 Aug 2026 00:00:00 +0700

This study aims to optimize the clustering of poverty-level data in Indonesia using the K-Means clustering method with a Z-Score normalization approach and evaluation using the Davies-Bouldin Index (DBI). The main problem in clustering poverty data is the difference in scales among variables, which can affect the clustering results. Therefore, data normalization using the Z-Score method was performed to equalize the scales among variables. The data used in this study are secondary data consisting of 514 regencies/cities in Indonesia entitled “Classification of Poverty Levels in Indonesia,” obtained from the Kaggle repository. The dataset was compiled by Ermila, Marcelo, and Faisal Dino Bahtiar and was last updated three years ago, specifically in 2023, as stated on the website. The data consist of poverty indicators that have undergone a variable selection process, in which variables that are irrelevant or have similar meanings and may potentially cause bias were excluded from the analysis. The clustering process was conducted using two scenarios, namely K=2 and K=3, to compare the clustering results. The results show that the K-Means method can effectively cluster the data based on poverty-level categories. The evaluation using the Davies-Bouldin Index shows DBI values of 0.9417 for K=2 and 2.706 for K=3. Based on the comparison of the clustering results, the K=2 scenario produced a lower DBI value and was therefore used as a reference for determining the classification levels. The contribution of this study lies in the development of a clustering process that integrates indicator selection, Z-Score normalization, the K-Means algorithm, and Davies-Bouldin Index evaluation across 514 regencies and municipalities in Indonesia to obtain regional groupings based on similarities in the characteristics of poverty indicators. Thus, the approach developed in this study can produce more optimal clustering results and can be used as a basis for supporting decision-making in poverty alleviation efforts effectively and appropriately.

Sistem Rekomendasi Lowongan dan Pencarian Talent Alumni Menggunakan Content-Based Filtering dengan Weighted Feature Matching

Muhammad Ihsan Nurulhuda — Mon, 31 Aug 2026 00:00:00 +0700

The development of information technology encourages the need for systems that are capable of managing alumni data while supporting the utilization of such data for career development purposes. In IKATEKSI UII, alumni data management and membership services were previously not integrated into a single platform, resulting in limited utilization of alumni data for career information and talent searching needs. This study aims to develop a web-based alumni management and membership platform by implementing the Content-Based Filtering (CBF) method using the Weighted Feature Matching approach on job recommendation and Talent Search features. This study integrates alumni data management and membership services with an attribute-based matching mechanism using Content-Based Filtering and Weighted Feature Matching to support job vacancy recommendations and talent searches based on profile suitability. The system development method used is the Waterfall model, consisting of analysis, design, implementation, and testing stages. The system was developed using the Laravel framework with MySQL as the database. The results show that the developed platform successfully integrates alumni data management, membership services, and career development features based on profile attribute matching. The evaluation of the CBF method on the Talent Search feature achieved a Precision value of 1.000, Recall of 0.706, and F1-Score of 0.828. In addition, Black Box Testing results indicate that all main system functions operate according to the designed requirements.

Prediksi Arus Hubung Singkat Menggunakan Model Regresi Struktural dan Evaluasi Dampaknya terhadap Rating Peralatan Gardu Induk 500 kV

Bryan Bima Laksana, Tri Wahyudi Adi — Mon, 31 Aug 2026 00:00:00 +0700

The development of power systems and the strengthening of transmission networks can increase short-circuit current levels, potentially reducing equipment safety margins when fault currents approach or exceed the rated breaking capacity. This study aims to identify the factors influencing short-circuit current variations, analyze the relationships among the relevant variables, predict short-circuit current development for the 2026–2030 period, and evaluate the adequacy of 500 kV substation equipment ratings. The proposed solution is an integrated approach that combines data-driven short-circuit current prediction with technical evaluation of equipment ratings, allowing potential increases in fault current levels to be identified before they exceed the interruption capability of installed equipment. The study employs Partial Least Squares Structural Equation Modeling (PLS-SEM) using historical power system data from 2018–2025. The analytical procedure includes the development of the Generation, Network Configuration, System, and Short Circuit constructs, evaluation of the measurement model using outer loading, Composite Reliability, Average Variance Extracted (AVE), and discriminant validity, followed by structural model evaluation using path coefficients, bootstrapping, t-statistics, and p-values. The validated model is subsequently used to generate latent variable scores and project short-circuit current development through 2030. The results indicate that the System variable has a positive and statistically significant effect on Short Circuit, with a path coefficient of 0.173, t-statistic of 2.156, and p-value of 0.016, whereas the direct effects of Generation and Network Configuration on Short Circuit are not statistically significant. The projection results indicate an increasing trend in short-circuit current at several observation points. Further equipment assessment shows that 19 extra-high-voltage substations have predicted short-circuit current levels exceeding the breaking capacity of their installed equipment, indicating the need for further attention in equipment reinforcement and system reliability planning.

Analisis Audit Energi Mekanikal dan Elektrikal Berbasis Significant Energy Use (SEU) pada PLTU Stoker

Wisnu Yoga Prasetya, Dhany Harmeidy Barus — Mon, 31 Aug 2026 00:00:00 +0700

Energy audits are a crucial strategic instrument for optimizing power plant operational efficiency while supporting national energy conservation programs. This study aims to evaluate the performance of mechanical and electrical systems based on the Significant Energy Use (SEU) approach, as well as to identify applicable energy-saving opportunities in a 7 MW stoker-type coal fired power plant. The research method employed is a detailed energy audit approach through thermal performance calculations using heat rate analysis (input-output and heat loss methods), evaluation of the heat rate gap against the historical baseline, and mapping of Significant Energy Use (SEU) across all main mechanical and electrical equipment. The energy audit results reveal a significant performance degradation, where the actual High Pressure Turbine (HPT) efficiency only reaches 46.52% and the condenser vacuum pressure decreases to 180.32 mmHgA. In addition, auxiliary power consumption is recorded at a relatively high level of 11%. Through operational interventions, the largest energy-saving opportunities can be achieved by restoring the condenser vacuum pressure and improving HPT efficiency. Overall, the total identified energy-saving opportunities amount to 71,765.11 GJ/year, equivalent to 1,714 TOE/year, with an estimated economic value reaching IDR 5,077,394,462.83 annually. Furthermore, reducing auxiliary power from 11.00% to 9.52% provides a potential saving of 50.95 kCal/kWh, alongside savings of 221.80 kCal/kWh from condenser vacuum pressure improvement and 167.49 kCal/kWh from HPT efficiency enhancement. The conclusion of this study confirms that the periodic implementation of technical recommendations derived from the audit is highly effective in reducing the Net Plant Heat Rate (NPHR), lowering the basic cost of electricity provision and ensuring sustainable compliance with environmental regulations.