Jurnal Pilar Nusa Mandiri

UTILIZING END USER DEVELOPMENT METHOD FOR DEVELOPING PENCAK SILAT ORGANIZATION INFORMATION SYSTEMS

Heribertus Ary Setyadi — 2025-09-23

Gondang is one of the PSHT sub-branches located in Sragen Regency, Central Java, Indonesia. In managing member data from recruitment to promotion, conventional methods are still used using office applications and information dissemination is still using brochures and social media. This research aims to develop an information system that can help manage data and disseminate information at PSHT Gondang. The system developed can manage the registration of prospective member to become a member and the process of promotion. Delivery of information in the form of organizational structures, announcements, activity schedules, services for member and community, activity galleries containing photos and videos can also be accessed through the system.EUD was chosen as a method in system development because time required is quite short with a relatively small cost allocation. The system is created using Laravel framework and Firebase as a database with a responsive display so that it can be accessed using a smartphone. By using the EUD method, users can modify the appearance and existing information if there is a change in data from the organization which was not available in previous research.

COMPARISON OF ARIMA, LSTM, AND GRU MODELS FOR FORECASTING SALES OF HIT AEROSOL PRODUCTS

Nendi Sunendar — 2025-09-23

A more accurate forecasting model, such as LSTM, can significantly enhance business efficiency by providing more reliable predictions of future sales, allowing for better inventory management, optimized production schedules, and more precise distribution planning. This leads to reduced costs, minimized stockouts, and improved customer satisfaction. This study evaluates the forecasting performance of ARIMA, Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) models using sales data from 2021 to 2023. The models are assessed based on Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). Results show that LSTM outperforms the other models with a MAPE of 10.76%, followed by ARIMA at 11.23% and GRU at 11.47%. These findings highlight the advantages of deep learning methods, particularly LSTM, in capturing complex patterns and trends in time series data. The study demonstrates the potential of these models to optimize sales forecasting, aiding decision-making processes in production and distribution planning.

KOPTIHUB: A WAREHOUSE APPLICATION PROTOTYPE FROM COOPERATIVE PERS PECTIVE

Luh Made Wisnu Satyaninggrat — 2025-09-23

Effective warehouse management is crucial for ensuring the availability of raw materials and smooth product distribution, particularly at Sentra Industri Kecil Somber (SIKS) Balikpapan, which specializes in soybean-based industries. Manual record-keeping has presented significant challenges, leading to recording errors, stock discrepancies, and delays in raw material procurement. To address these issues, a digital warehouse management prototype, "KoptiHub," was developed using a User-Centered Design (UCD) approach. This approach aimed to enhance inventory tracking efficiency, streamline raw material ordering, and improve overall product distribution. The prototype was evaluated using the System Usability Scale (SUS) with 15 cooperative administrators at SIKS Balikpapan. The evaluation yielded an SUS score of 82.17, resulting in an "A" grade, which indicates high usability and strong alignment with user expectations. Compared to previous warehouse management solutions, KoptiHub demonstrates superior usability, particularly in cooperative settings. However, further improvements, such as a simplified user interface and an AI-driven inventory forecasting feature, could enhance efficiency and accessibility. The results suggest that KoptiHub could serve as a scalable model for digitizing warehouse management in MSMEs and cooperatives, aligning with emerging trends in smart inventory management and supply chain optimization.

WASTEWISE: AI-POWERED WASTE EDUCATION FOR ELEMENTARY STUDENTS USING YOLOV8 AND ESP32-CAM

Kenny Aldi — 2025-09-23

The growing volume of global waste poses a significant challenge for effective waste management, particularly in developing countries where awareness and practices around waste sorting remain limited. This study aims to enhance elementary school students' understanding and efficiency in sorting organic and inorganic waste using an interactive, AI-powered educational tool. The proposed system, WasteWise, integrates YOLOv8 for real-time object detection and ESP32-CAM for capturing waste images. A pre-test and post-test experimental design was conducted to assess students’ performance before and after using the system. The results showed a notable improvement in sorting accuracy, increasing from 60% with manual sorting to 90% using the WasteWise system, alongside reduced sorting time. These findings highlight the system's potential not only as an automated waste classification tool but also as a cost-effective and engaging platform for promoting environmental awareness and digital literacy among young learners.

SENTIMENT ANALYSIS ON TRAINING IMPLEMENTATION’S FEEDBACK IN PT XYZ

Fadilia Rinarwastu — 2025-09-23

Customer satisfaction is an important aspect in building a company's image, both for employees and external parties. In order to improve employee satisfaction and performance, training that organized by the company needs to receive feedback so that the training organizers can continue to provide the best service to employees who participate in the training. The large volume of feedback that must be processed in text form, leads to prolonged identification of comments and the omission of certain training programs from further analysis. This study applies text mining using sentiment analysis and Word Cloud visualization to evaluate the effectiveness of training methods and identify areas for improvement based on employee feedback on training programs at PT XYZ. The amount of data used after preprocessing was 48,910 open feedback responses from 4,314 training sessions consisting of three forms: classroom training, digital learning, and hybrid learning. The evaluation for clustering used the K-Means method, which turned out to use two optimal clusters based on the silhouette. Overall satisfaction with the training was determined through key points such as stable internet connection, overlapping of training schedule, and poor learning environment. Issues frequently that identified in the Word Cloud analysis revealed keywords describing positive and negative aspects of the situation that are requiring further improvement. This identification is useful for developing recommendations to enhance the implementation of the training and participants' experience. Further research may also involve advanced sentiment analysis and more accurate classification methods.

ISOLATION FOREST PARAMETER TUNING FOR MOBILE APP ANOMALY DETECTION BASED ON PERMISSION REQUESTS

Valencia Claudia Jennifer Kaunang — 2025-09-23

Ensuring mobile app security needs the capability to detect apps that request excessive or inappropriate permissions. This research proposes an anomaly detection approach using Isolation Forest, enhanced through hyperparameter tuning, to identify suspect apps based on permission request patterns. The dataset is processed into binary features, followed by exploratory data analysis (EDA) to examine the distribution and highlight sensitive permissions. The Isolation Forest model is then optimized by tuning parameters such as contamination level, number of estimators, and sample size. The fine-tuned model achieved a more accurate separation between normal and anomaly applications, detecting 10 anomalies out of 200 applications, with anomaly applications averaging 125.10 permits compared to 42.76 in normal applications. These anomalies often requested permissions related to network, storage, contacts and microphone, indicating potential privacy risks. The results show that parameter tuning improves the detection performance of Isolation Forest, providing a practical solution for mobile security monitoring. After tuning, the number of false positives decreased by 50%, and the model successfully reduced detected anomalies from 20 to 10, increasing the precision of anomaly detection from 70% to 90%. Future work could include improving feature selection and integration into real-time detection systems.

APPLICATION OF ARTIFICIAL NEURAL NETWORK METHODS TO DETECT HEART ATTACKS

Nasir Hamzah — 2025-09-23

A heart attack is a medical emergency caused by restricted blood flow to the heart, commonly leading to myocardial infarction due to blood clots or fat accumulation. Early detection of heart disease is crucial to support prevention efforts and assist healthcare professionals in timely diagnosis and treatment. This study applies the Backpropagation Neural Network (BPNN) algorithm as an intelligent computing method for heart attack detection. Experimental results demonstrate a prediction accuracy of 96.47%, confirming the effectiveness of artificial neural networks in identifying heart attacks in patients. These findings highlight the potential of BPNN as a reliable and precise early detection system, which can support more accurate clinical decision-making and improve the effectiveness of heart attack prevention and treatment.

IDENTIFICATION OF FOOD DIVERSIFICATION ON JAVA ISLAND USING ARCGIS

Amir Murtako — 2025-09-23

Indonesia is addressing the challenges of food security and consumer preference also known as Food diversification. The research aims to analyze the potential of various local food sources as alternatives to rice, which is the dominant staple food in Indonesia, with a particular focus on geographic implications. Although local carbohydrate sources like corn, potatoes, and tubers are available, their adoption is limited and understudied in relation to geographic distribution and consumer behavior. This study integrates survey data and GIS-based spatial analysis to evaluate local food diversification potential. Findings show that while 100% of respondents consume rice, 48.7% have tried alternatives, with limited availability (41.03%) and higher costs (17.95%) as key barriers. With 94.7% expressing willingness to adopt new staples, the results suggest GIS-based decision support systems can guide effective, region-specific food policy interventions.

EVALUATING PREPROCESSING EFFECTS IN NAME RETRIEVAL USING CLASSICAL IR AND CNN-BASED MODELS

Frizca Fellicita Marcelly — 2025-09-23

Information Retrieval (IR) systems are pivotal for efficient data management, particularly in tasks involving name searches and entity identification. This study evaluates text preprocessing techniques, including case folding, phonetic normalization, and gender tagging, that affect the performance of classical (TF-IDF, LSI) and CNN-based retrieval models for multilingual name matching. Using a dataset of 365,468 globally diverse names, this study implements a preprocessing pipeline featuring: Double Metaphone phonetic preprocessing (92% validation accuracy), gender disambiguation for unisex names (92% accuracy), and optimized n-gram tokenization for short names. Evaluation metrics include precision, recall, F1-score, and our novel Name Similarity Score (NSS), combining orthographic and phonetic preprocessing. Results show our full pipeline improves recall to 1.00 and F1-score by 37% while reducing false negatives by 63%. Key findings reveal: TF-IDF achieves superior recall (0.98 vs CNN’s 0.85), LSI handles cultural variants effectively, and CNNs deliver the highest precision (0.91 vs TF-IDF’s 0.70), particularly for unisex names. This work contributes both a scalable multilingual preprocessing framework and the NSS evaluation metric for robust name retrieval systems.

COMPARATIVE PERFORMANCE OF TRANSFORMER AND LSTM MODELS FOR INDONESIAN INFORMATION RETRIEVAL WITH INDOBERT

Nendi Sunendar Sunendar — 2025-09-23

Neural network-based Information Retrieval (IR), particularly with Transformer models, has gained prominence in information search technology. However, the application of this technology in Indonesian, a low-resource language, remains limited. This study aims to compare the performance of the LSTM model and IndoBERT for IR tasks in Indonesian. The dataset consists of 5,000 query–document pairs collected via scraping from three Indonesian news portals: CNN Indonesia, Kompas, and Detik. Evaluation was performed using MAP, MRR, Precision@5, and Recall@5 metrics. The results show that IndoBERT outperforms LSTM in all metrics with a MAP of 0.82 and MRR of 0.84, while LSTM only reached a MAP of 0.63 and MRR of 0.65. These findings confirm that Transformer models like IndoBERT are more effective at capturing semantic relevance between queries and documents, even with limited datasets.

PERCEPTION AND BARRIERS TO MOOC ADOPTION: A CASE STUDY OF KARTU PRAKERJA RECIPIENTS

Dendy Herdianto — 2025-09-23

The Indonesian government launched the Pre-Employment Card (Kartu Prakerja) program to enhance workforce skills and address economic challenges. This program provides training through online platforms, including Massive Open Online Courses (MOOCs). The UTAUT2 model was employed as a framework to understand the factors influencing the acceptance and use of educational technology in this context. This study examines the effects of UTAUT2 variables—performance expectancy, effort expectancy, habit, traditional barriers, platform content, access limitations, interaction limitations, facilitating conditions, hedonic value, price value, and social influence—on the intention and adoption of MOOCs among Pre-Employment Card participants. The sample consisted of 222 respondents who were users of the Prakerja platform. Data were collected using a questionnaire and analyzed through Structural Equation Modeling (SEM) with the support of PLS-SEM software. In addition, a sentiment analysis was conducted on comments posted on the official Instagram account @prakerja.go.id to explore public perceptions of the program. The findings reveal that 46.2 percent of public sentiment was negative, particularly related to the program implementation and the use of partner MOOC platforms. SEM analysis further indicates that hedonic value, habit, and social influence have positive and significant effects on the intention and adoption of MOOCs. The moderation analysis by gender shows that performance expectancy, hedonic value, and social influence are stronger among males, whereas effort expectancy, habit, and platform content are stronger among females.

COMPARATIVE ANALYSIS OF RANDOM FOREST AND SUPPORT VECTOR CLASSIFIER FOR PREDICTING STUDENTS’ ON-TIME GRADUATION

Nurus Sarifatul Ngaeni — 2025-09-23

On-time graduation is one of the key indicators of educational quality in higher education. The influencing factors range from students’ internal issues and academic abilities to institutional policies. However, academic management has not yet been able to classify the data and analyze the underlying factors contributing to delayed graduation. By identifying these factors, management can formulate appropriate academic solutions or policies. The purpose of this study is to build a prediction model for on-time graduation using machine learning algorithms. This study compares the classification performance of the Random Forest algorithm and the Support Vector Classifier (SVC). The dataset, consisting of 1,298 student records, includes academic data such as study program, GPA, TOEFL score, cohort year, and study duration. Model performance was evaluated using accuracy, F1 score, and ROC-AUC metrics, followed by a confusion matrix analysis. The final evaluation revealed that the Random Forest algorithm achieved the best performance, with an accuracy of 91.86%, an F1 score of 91.86%, and a ROC-AUC of 97.39%. Meanwhile, the SVC model obtained an accuracy of 81.12% and an F1 score of 81.09%. Based on these results, it can be concluded that the Random Forest algorithm is more reliable as a prediction model in the academic domain. The main contribution of this study is the development of an early detection system for students at risk of delayed graduation. Furthermore, the findings can serve as a basis for designing more solution-oriented academic policies in accordance with the conditions at STIMIK Tunas Bangsa Banjarnegara.

ADDRESSING DIGITAL STARTUP FAILURE THROUGH THE AGILE METHODOLOGY APPROACH: A SYSTEMATIC LITERATURE REVIEW

Kenedi Binowo — 2025-09-23

Startups are recognized as emerging enterprises that contribute to job creation, economic stabilization, and national development. Digital startups are formed to address challenges within their environments. This study aims to provide solutions and preventive measures for digital startup failures, given the persistently high global failure rate of 90%. A systematic literature review (SLR) was conducted to identify Agile-based Critical Success Factors (CSFs), which were then mapped as solutions to mitigate digital startup failures. Based on the findings, the most significant contributing factor to the failure of digital startups is insufficient funding (i.e., running out of capital or financial resources). To address this issue, the agile method offers relevant solutions that can be mapped to the problem, namely the adoption of “Iterative Budget Management,” “Accurate Effort Estimation,” and “Risk Management Strategies.” This study provides practitioners with valuable insights, knowledge, and reference points regarding the critical success factors (CSFs) derived from agile practices, which can serve as strategic mechanisms for mitigating failure in early-stage startups. Moreover, the research is expected to contribute new theoretical understanding that informs potential solutions to prevent digital startup failure.

MACHINE LEARNING FOR EMPLOYMENT POSITION MAPPING

Sena Aditia Apriadi — 2025-09-23

Employee performance directly impacts organizational efficiency, yet traditional HR analytics often lack predictive precision. This study bridges HR theory and machine learning by evaluating tree-based algorithms for employee data analysis. Using a dataset of 15,227 employee records, we tested the Bagged Decision Tree algorithm, focusing on variables such as talent, career values, and aspirations. The Bagged Decision Tree achieved 98.65% accuracy, with talent and career values as key predictors. Excluding aspiration values reduced accuracy slightly to 98.57%, while excluding career values lowered it significantly to 92.13%. These findings highlight the robustness of the Bagged Decision Tree in HR analytics and emphasize the importance of variable selection, particularly career values and talent, in predicting performance outcomes. Future work should further explore real-world implementation challenges.

ANALYSIS OF THE NEED FOR AN INFORMATION SYSTEM ON PRICES AND AVAILABILITY OF BASIC MATERIALS

Andriyan Dwi Putra — 2025-09-29

The development of information technology has driven digital transformation in various sectors, including the economic sector. Managing data on the prices and availability of basic commodities is crucial for maintaining community economic resilience. This study applies a design thinking approach to analyze the need for an information system on the prices and availability of basic commodities in Yogyakarta City, with a testing plan prepared using black box, white box, and security methods. The analysis produced three main findings: the need for Single Sign-On (SSO) with role-based access, real-time monitoring of commodity prices, and cross-agency integration in agenda and program management. The proposed system design consists of four main modules: administration, agenda, services, and programs/activities. Since this study is limited to the needs analysis and prototype design stage, empirical test results are not yet available. Nevertheless, the study provides an initial framework and foundation for cross-agency integration in the Yogyakarta City Government to support transparency, coordination, and control of basic commodity prices.

PADANG FOOD IMAGE CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORK (CNN)

Nabilah Putri Permana — 2025-09-29

The recognition of Padang traditional foods presents a challenge because of their high visual similarity, which makes manual classification difficult. This study aims to develop an automatic image classification model for Padang foods using the Convolutional Neural Network (CNN) algorithm. The dataset consisted of 1350 images across nine classes of Padang dishes including omelet, chili egg, cow tendon curry, stuffed intestine curry, fish curry, dendeng batokok, rendang, ayam pop, and fried chicken. The CNN architecture was trained for twenty epochs and evaluated using accuracy, loss, confusion matrix, and testing with new images. The results show that the model reached a final training accuracy of 70.2 percent and a validation accuracy of 65 percent, while testing with unseen images produced correct predictions with moderate confidence levels. These findings suggest that CNN is effective for classifying Padang traditional foods and can be applied in culinary promotion, digital food catalogs, and technology based ordering platforms.