Journal of Intelligent Decision Support System (IDSS)

Naïve bayes classification for oil palm leaf disease based on color and texture features

Alvin Kesuma, Natasya Ate Malem Bangun, Meida Cahyo Untoro — Tue, 30 Sep 2025 00:00:00 +0000

This study presents a comparison between standard Naïve Bayes classifier and its Genetic Algorithm-optimized variant for automated classification of oil palm leaf diseases. The system incorporates RGB color features alongside texture features extracted using the Gray Level Co-Occurrence Matrix. A dataset of of 225 JPG images of oil palm leaves, divided into training and testing sets in an 80:20 split is used. The methodology consisted of preprocessing, feature extraction, and classification. In the preprocessing phase, images were manually cropped, resized to 256 × 256 pixels, and background elements were removed. Feature extraction was then performed to obtain RGB color values and GLCM-based texture values, including contrast, correlation, energy, and homogeneity. Classification was conducted using two variants of the Naïve Bayes algorithm: one with default parameters and another optimized via GA for the Laplace smoothing hyperparameter. Model performance was assessed using a confusion matrix, with accuracy, precision, and recall serving as the primary evaluation metrics. Experimental results showed that both models achieved identical performance, with an accuracy of 51%, a precision of 52%, and a recall of 51%. These findings suggest that the Naïve Bayes classifier, even in its baseline form, demonstrates low discriminative performance for oil palm leaf disease detection, and when enhanced through GA-based optimization, it still provides only limited effectiveness. Therefore, this research highlights the need to pursue alternative methodologies, such as deep learning techniques or the adoption of more discriminative feature representations, aimed at improving both the accuracy and robustness of image-based disease detection in agriculture.

Naïve bayes algorithm for early diagnosis of non-communicable diseases

Nuzul Ramadhan, Rakhmat Kurniawan Ramli — Tue, 30 Sep 2025 00:00:00 +0000

Non-communicable diseases such as heart disease, diabetes mellitus, hypertension, stroke, asthma, rheumatism, and GRED are still the main causes of illness and death in Indonesia. This problem is more serious in rural areas with limited health services, such as Lubuk Palas Village, Asahan Regency, which faces obstacles in distance, road infrastructure, and a limited number of medical personnel, so early diagnosis is often neglected. This research aims to apply the Naïve Bayes method in a non-communicable disease diagnosis expert system and develop web and mobile-based applications to support the community and medical personnel in early detection. The research method combines primary data from observations and interviews with health workers and secondary data from medical literature. Each symptom is given a probability weight of 0.00–1.00 according to medical consultation, then processed using the Naïve Bayes algorithm with two approaches, namely direct calculation and gradual filtering. The results show that the system produces a posterior probability of 99.32% in the heart disease scenario with typical symptoms and 90.00% in the stroke scenario with partial symptoms. The findings of this research are that the application of two Naïve Bayes inference pathways is proven effective in producing an initial diagnosis that is adaptive to variations in symptoms, relevant for rural conditions with limited health services, and capable of providing fast, practical, and widely accessible medical decision support.

Web-based application development for the digitalization of badminton court reservation and scheduling using scrum methodology

Arief Ichwani, Mohamad Idris, Aidil Afriansyah — Tue, 30 Sep 2025 00:00:00 +0000

The primary challenge in managing badminton court reservations and scheduling lies in the absence of an integrated system that supports booking, payment flexibility, and efficient schedule management. This study addresses these issues by developing a badminton court reservation information system equipped with user registration, login, court booking, schedule management, payment reporting, and member schedule search features. System evaluation was conducted through a User Acceptance Test (UAT), which confirmed that all core features functioned effectively and met user requirements. To further assess user experience, a Mean Opinion Score (MOS) evaluation involving five respondents and ten questions was carried out, yielding an average score of 3.8 on a four-point scale, categorized as very good. Respondents indicated that the system is easy to use, offers intuitive navigation, has an attractive interface, and provides stable performance while accelerating the booking process. These findings demonstrate that the system is both functionally reliable and well-received by users, thereby contributing to improved administrative efficiency, payment transparency, and overall user convenience.

Expert system for diagnosing EDC cash register malfunctions using the Decision Tree method

Novi Thalia, Akim Manaor Hara Pardede, Husnul Khair — Tue, 30 Sep 2025 00:00:00 +0000

Problems with the use of Electronic Data Capture (EDC) machines at Jinjja Chicken Center Point Medan restaurant pose a significant challenge, especially since EDC machines not only function as a means of cashless payment, but also as part of the cashier's operational system. Frequent disruptions include program errors, display errors, total EDC shutdowns, line idles, and “please try again” messages. Until now, the process of reporting and repairing EDC malfunctions has been done manually by submitting a request to the bank, which often makes the diagnosis and repair process slow and inefficient. This is exacerbated by the limited technical information available to restaurant operators when customers experience disruptions. To overcome these problems, this study aims to develop an expert system for diagnosing EDC cash register malfunctions using the Decision Tree method, which is capable of mimicking the way an expert diagnoses EDC problems quickly and accurately. The Decision Tree method was chosen because it is capable of mapping the decision-making process based on attributes or symptoms that arise, to produce a conclusion in the form of the type of malfunction. This system was built using the PHP programming language and run locally using XAMPP as a web server. The research was conducted in a limited setting at the Jinjja Chicken Center Point restaurant in Medan, with five main malfunction categories as variables: Program Error, Display Error, EDC Completely Dead, Line Idle, and Please Try Again. The final result of this system development is expected to provide practical, efficient solutions that approximate the capabilities of an expert, as well as make a real contribution to the utilization of expert system technology to assist in the diagnosis of digital device damage in the service sector.

Tidal flood prediction in Indonesian coastal areas using long short-term memory for enhanced early warning systems

Agung Mahadi Putra Perdana — Tue, 30 Sep 2025 00:00:00 +0000

Coastal flooding, locally known as banjir rob, persists as a recurring hazard in Indonesia’s low-lying coastal zones, driven by tidal variation, river discharge, and meteorological dynamics. This study applies a Long Short-Term Memory (LSTM) neural network for short-term flood prediction using a multivariate dataset covering 2020–2024. The dataset integrates daily records of water levels from six monitoring stations (Katulampa, Pos Depok, Manggarai, Istiqlal, Jembatan Merah, Flusing Ancol), sea-level observations from Marina Ancol, and meteorological parameters including wind speed, wind direction, rainfall, atmospheric pressure, and sea surface temperature. Flood status was encoded as a binary target (0 = non-flood, 1 = flood) with balanced distribution, enabling robust model generalization. Preprocessing involved data cleaning, normalization, and sliding-window sequencing to capture temporal dependencies. The LSTM architecture combined stacked recurrent layers, dropout regularization, and a dense output layer, trained in TensorFlow with tuned hyperparameters. Evaluation indicated strong predictive skill, with Mean Absolute Error (MAE) below 3 cm, Mean Absolute Percentage Error (MAPE) under 2%, and classification accuracy above 90%. Comparative analysis demonstrated consistent outperformance of LSTM over Artificial Neural Networks (ANN) and linear regression, both of which produced higher errors and weaker representation of temporal patterns. The findings confirm LSTM’s capacity to support operational early warning systems, strengthen community preparedness, and mitigate socio-economic impacts in vulnerable coastal regions.

Real-Time Information System Integration In Coal Production And Distribution Management

Tue, 30 Sep 2025 00:00:00 +0000

Coal production faces challenges in data management and operational coordination efficiency. This study examines the integration of an information system using AI and real-time capabilities to improve the efficiency and effectiveness of coal production. With a system development approach using the SDLC Waterfall method and dummy data simulation, the system successfully accelerates production monitoring, increases data accuracy, and supports rapid decision-making. These findings provide a strong foundation for developing adaptive and effective mining information systems in the digital era. The applied method delivers optimal results in measuring pressure and temperature as well as detecting coal sensor values using an AI-based IoT system. Evaluation results show a reduction in reporting response time by up to 40%, an increase in stored data accuracy up to 95%, and positive feedback from simulation users regarding ease of access and production monitoring.

Development of interactive E-Modules on entrepreneurship based on augmented reality for students of SMKN 6 Medan

Tue, 30 Sep 2025 00:00:00 +0000

Entrepreneurship education in vocational high schools (SMK) is often dominated by conventional methods that make students passive and less able to connect theory with practice, especially on abstract topics such as franchising. This study aims to develop an Augmented Reality (AR)-based entrepreneurship e-Module that is interactive, contextual, and oriented toward vocational learning. Using a Research and Development (R&D) approach with the ADDIE model (Analysis, Design, Development, Implementation, Evaluation), the study involved 100 grade XI students at SMK Negeri 6 Medan. Data were collected through expert validation, questionnaires, observations, interviews, and pre- and post-tests. The validation results placed the e-Module in the highly valid category (>80%), while the N-gain test showed a significant increase in the experimental class (0.83; high category) compared to the control class (0.32; low category). The t-test also confirmed a significant difference (p < 0.05) between the two groups. In addition, observations indicated improvements in student motivation, engagement, and understanding, as well as greater ease for teachers in delivering abstract material. These findings demonstrate that AR integration in entrepreneurship e-Modules can effectively bridge theory and practice, enhance learning quality in vocational schools, and support the achievement of SDG 4 on quality education.

Comparison of algorithm performance, Random Forest Regression, SVR, and Gradient Boosting in predicting academic grades based on student lifestyle

Danny sihombing — Tue, 30 Sep 2025 00:00:00 +0000

This study examines the effectiveness of three machine learning algorithms—Random Forest Regression, Support Vector Regression, and Gradient Boosting—in predicting students’ academic grades based on lifestyle-related factors including study hours, sleep duration, social interaction, physical activity, and stress levels. Employing a quantitative experimental approach, model performance was evaluated using R², MSE, RMSE, and MAE, while SHAP analysis was applied to interpret feature importance. The results show that all models achieved reasonable predictive accuracy, with Gradient Boosting consistently outperforming the others across all metrics. Study duration was identified as the most influential predictor, whereas stress level and gender had minimal impact. These findings emphasize the importance of non-academic lifestyle factors in predicting academic achievement and provide insights for the development of data-driven, personalized decision support systems in education.