Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi)
Not a member yet
1071 research outputs found
Sort by
Comparison of Sugarcane Drought Stress Based on Climatology Data using Machine Learning Regression Model in East Java
No full textCrop Water Stress Index (CWSI), derived from vegetation features (NDVI) and canopy thermal temperature (LST), is an effective method to evaluate sugarcane sensitivity to drought using satellite data. However, obtaining CWSI values is complicated. This study introduces a novel approach to estimate CWSI using climatological data, including average air temperature, humidity, rainfall, sunshine duration, and wind speed features obtained from the local weather station BMKG Malang City, East Java, for the period 2021-2023. Before estimating CWSI, we analyzed sugarcane water stress phenology, examined the strength of the correlation between climatological features and CWSI, and looked at the potential for adding lag features. Our proposed prediction model uses climatological features with additional Lag features in a machine learning regression approach and 5-fold cross-validation of the training-testing data split with the help of optimization using hyperparameters. Different machine learning regression models are implemented and compared. The evaluation results showed that the prediction performance of the SVR model achieved the best accuracy with R2 = 90.45% and MAPE = 9.55%, which outperformed other models. These findings indicate that climatological features with lag effects can effectively predict water stress conditions in rainfed sugarcane if using an appropriate prediction model. The main contribution of this study is the utilization of local climatological data, which is easier to obtain and collect than sophisticated satellite data, to estimate CWSI. The application of the results shows that climatological data with lag effects can accurately estimate water stress conditions in rainfed sugarcane. In drought-prone areas, this strategy can help sugarcane farmers make better choices about land management and irrigation
Strategic Approach to Enhance Information Security Awareness at ABC Agency
No full textInformation security awareness (ISA) is crucial to an organization's cybersecurity strategy, particularly since employees are often the last defense against cyberattacks. Despite regular communication on cybersecurity threats, the ABC Agency has not evaluated the level of ISA among its employees, leaving a gap in understanding the effectiveness of its awareness programs. This is critical, as the agency handles highly confidential data that could be at risk of accidental or intentional leaks. The Kruger Approach and the Human Aspect of Information Security Questionnaire (HAIS-Q) were used in this study to measure the ISA levels of employees at the ABC Agency. We employed the Analytic Hierarchy Process (AHP) method to analyze data collected from 86 respondents. The findings indicate that ABC Agency employees demonstrate satisfactory ISA overall. However, the "Internet Use" dimension received a medium rating, underscoring the necessity for focused enhancements in this domain. These results underscore the importance of tailoring information security awareness programs to address specific weaknesses. We provide strategic recommendations to enhance the agency's cybersecurity posture. Furthermore, this study opens avenues for future research on ISA measurement across various public and private organizations
Eye Disease Detection and Classification Optimization Using EfficientNet-B5 with Emphasis on Data Augmentation and Fine-Tuning
No full textEye diseases such as glaucoma, cataract, and diabetic retinopathy pose significant global health challenges, underscoring the need for accurate and efficient diagnostic systems. This study employed the EfficientNet-B5 model to enhance the detection and classification of eye diseases by incorporating advanced data augmentation and fine-tuning techniques. The research utilizes the Ocular Disease Intelligent Recognition (ODIR) dataset, consisting of 4,217 fundus images categorized into four classes: normal, glaucoma, cataract, and diabetic retinopathy. The methodology comprises three phases: baseline model training, model training with data augmentation, and fine-tuning. The baseline model achieved an accuracy of 60.43%, which improved to 63.03% with data augmentation—an increase of 2.6 percentage points. Fine-tuning further elevated the accuracy to 93.23%, representing a notable improvement of 33.8 percentage points over the baseline. Model performance was evaluated using standard classification metrics including accuracy, precision, recall, and F1-score. These findings demonstrate the technical efficacy of combining augmentation and fine-tuning to enhance model generalization. The proposed approach offers a robust framework for developing dependable AI-driven diagnostic tools to support early detection and facilitate informed clinical decision-making
A Comparative Evaluation of YOLOv9 and DETR Models in Traffic Object Detection for Intelligent Surveillance Systems
No full textObject detection plays a crucial role in traffic surveillance, particularly in urban environments characterized by high vehicle density, diverse weather conditions, and limited computational resources. Although YOLOv9 and DETR have demonstrated strong performance in general object detection tasks, there is a lack of comparative research evaluating their effectiveness under specific challenges of traffic surveillance. These challenges include the need for real-time processing, accurate detection of small or partially occluded objects, and adaptability to complex traffic scenarios. This study addresses this gap by conducting a comparative evaluation of YOLOv9 and DETR using a custom traffic image dataset, with training iterations varied from 10 to 50 epochs to observe performance development. Evaluation metrics included mean average precision, precision, recall, F1-score, inference time, and object count per image. The results indicated that DETR achieved the highest accuracy across all metrics at the final training stage and detected up to 22 objects per image. However, the average inference time exceeded seven seconds per image, limiting the real-time applicability. Conversely, YOLOv9 achieved competitive accuracy with a significantly faster inference time of approximately 0.43 seconds per image. These findings provide practical insights into the trade-off between accuracy and processing efficiency, and offer guidance for model selection in operational traffic surveillance systems
A New Approach for Dynamic Analysis of Indonesian Food Prices using the PC Algorithm and Vector Autoregression
No full textFood prices are important global issue and their relationship with fuel prices has become a main concern in society. An increase in the subsidized fuel price on 3 September 2022 has allegedly caused a rise in food (grocery) prices. This paper conducts an empirical study to analyze the relationships between food prices in Indonesia: rice, chicken, beef, egg, red chili, cayenne, shallot, garlic, cooking oil, and sugar. The study uses time series data of food prices from 1 January 2018 to 31 December 2023, which consists of food prices from 87 traditional markets in Indonesia. The commodity prices are obtained from online public data provided by Bank Indonesia. It divides the analysis (pre- and post-3 September 2022) to see how the relationship between food prices changes due to the increase in the subsidized fuel price. It performs the Peter Clark (PC) algorithm to generate causal graphs from real datasets where the true graphs are unknown, complements the analysis by performing Vector Autoregression (VAR) to investigate the dynamic relationship between food prices, especially how the subsidized fuel price increase changes its dynamic relationship. The causal graphs from pre- and post-increasing fuel prices show the changes in the role of variable relationships, e.g., sugar and beef. The VAR results also show an interesting change in the IRF pattern. The results from both the PC algorithm and VAR show that there is a structural change in the relationship between food prices and that there is a different effect of price shock due to the subsidized fuel price increase. It might have been an indication of a change in the consumption pattern in society as a response to a food price increase. This must be a huge task to do in maintaining food prices when there is an adjustment in the subsidized fuel prices
Automated Ripeness Detection of Oil Palm Fruit Using a Hybrid GLCM-HSV-KNN Model
No full textAccurately determining the ripeness of oil palm fruit is crucial for ensuring the quality of palm oil. However, traditional manual methods are often time-consuming and less accurate. This study aimed to develop an automated system for detecting the ripeness of oil palm fruit by combining the Hue Saturation Value (HSV) model, Gray Level Co-occurrence Matrix (GLCM), and K-Nearest Neighbor (KNN) algorithms. This system utilizes K-Nearest Neighbors to classify the relationship between color features extracted using the HSV model and texture features derived from GLCM analysis to categorize fruit ripeness. The color features represent the fruit's chromatic characteristics associated with ripeness, while the texture features provide information regarding surface patterns related to ripeness. The color features represent the fruit's color characteristics associated with ripeness, whereas the texture features provide information about the surface patterns related to ripeness. The results indicate that the system can classify oil palm fruit into four distinct categories: Over-Ripe, Ripe, Half-Ripe, and Raw. The dataset was divided with an 80:20 ratio, where 80% was allocated for training data and the remaining 20% for test data. An accuracy rate of 85% was achieved. The results of this study demonstrate that the developed system effectively classifies oil palm fruit images based on ripeness levels. This system supports a sustainable automated palm oil production model through accurate ripeness detection, thereby reducing reliance on manual methods and enhancing consistency and productivity in palm oil processing. These findings indicate that the proposed hybrid method is feasible for integration into an automated classification system to support decision-making in oil palm harvestin
Benchmarking YOLOv8 Variants with Transfer Learning for Real-Time Detection and Classification of Road Cracks and Potholes
No full textRoad damage, including potholes and cracks, is a significant issue frequently encountered in road infrastructure in many regions. Such conditions accelerate road degradation, increase the risk of traffic accidents, and significantly increase the maintenance and repair costs. Although several deep learning models have been proposed for road damage detection, few studies have systematically compared the performance of lightweight YOLOv8 variants using a consistent dataset. To address this gap, this study proposes a road defect detection and classification model based on the YOLOv8 series, which is enhanced using transfer learning to improve performance and efficiency. The dataset, obtained from Roboflow, comprises 3,846 images categorized into training, validation, and testing sets. Three YOLOv8 variants—YOLOv8n, YOLOv8s, and YOLOv8m—were benchmarked for performance. A performance evaluation was performed using the metrics of precision, recall, and mean Average Precision (mAP). Results show that YOLOv8m achieved the highest precision (99.00%), recall (98.40%), and mAP (99.50%). In the pothole category, precision reached 98.70% and recall 99.30%; in the crack category, precision was 99.30% and recall 97.60%. The findings demonstrate that YOLOv8, particularly the YOLOv8m variant, is highly effective for real-time road damage detection and classification, offering a viable solution for intelligent transportation systems and automated infrastructure monitoring. This research has the potential to revolutionize infrastructure monitoring by enabling scalable, real-time, and cost-effective assessments of road conditions. It minimizes reliance on manual inspections, reduces human errors, and contributes to the development of intelligent transportation systems and predictive maintenance strategies
Improving Classification Performance on Imbalanced Stroke Datasets Using Oversampling Techniques
No full textStroke is the second leading cause of death worldwide and a major factor in long-term disability. Although early detection based on machine learning continues to be developed, it still faces challenges in the form of data imbalance, which can reduce classification performance. This study aimed to evaluate the effectiveness of several oversampling techniques, such as SMOTE, Borderline-SMOTE, and SVM-SMOTE, in improving the performance of stroke classification models on imbalanced data. The methods used included the application of three oversampling techniques, namely SMOTE, Borderline-SMOTE, and SVM-SMOTE, to balance the data distribution. Furthermore, Random Forest and XGBoost algorithms were used as classification models to identify stroke cases. The results of this study show that the use of oversampling techniques significantly improves model performance, especially in MCC and AUC metrics, compared to models without oversampling. Borderline-SMOTE provides the best results, with the highest accuracy of 96.45% on Random Forest and 96.41% on XGBoost, as well as MCC and AUC values that are consistently superior to other techniques. Furthermore, this study highlights that the use of Borderline-SMOTE significantly enhances model performance, as evidenced by an increase in MCC of 87.51% and an AUC of 45.40% in Random Forest, along with an increase in MCC of 76.52% and an AUC of 41.81% in XGBoost. These findings confirm that Borderline-SMOTE is an effective approach for dealing with data imbalance and improving the model's ability to detect minority classes in stroke classification
Benchmarking Machine Learning Paradigms for Resume Screening on Imbalanced Data
No full textManual resume screening is an inefficient and bias-prone process, yet comprehensive benchmarks of machine learning models on imbalanced, real-world recruitment data remain scarce. This study addresses this gap by benchmarking seven models from classical, ensemble, and deep learning paradigms for automated resume classification. Using a private dataset of 2,483 resumes across 24 job categories, this study evaluates the models with distinct TF-IDF and BERT embedding feature pipelines and an adaptive strategy for handling class imbalance (Class Weights, SMOTE, SMOTEENN). The results showed that the XGBoost model achieved the highest performance (weighted F1-score of 0.779), followed by the highly competitive BERT (F1 0.728) and Random Forest (F1 0.711) models. Despite these methods, all models struggled with extreme minority classes, confirming data scarcity as a primary limitation. This study provides a valuable benchmark and an evidence-based framework for HR practitioners, highlighting the critical trade-off between predictive performance (XGBoost), interpretability (Random Forest), and semantic capability (BERT). The findings conclude that the primary challenge is data representation, steering future work towards data augmentation and fairness audits
Software Product Line Engineering in Supply Chain Management Systems for Manufacturing Sector
No full textManufacturing companies are industrial enterprises that process raw materials and implement Supply Chain Management (SCM). SCM encompasses three stages: material management, planning and control, and production. While these stages are common across manufacturing companies, the workflows and strategies employed vary based on the type of goods produced. For example, one company typically approaches process orders based on requests, whereas the other processes orders based on stock availability. To address these similarities and differences, a software product line engineering (SPLE) approach can be utilized to develop SCM systems. This approach has already been proven effective in other cases, such as developing various product specifications for our Crowdfunding Application (Amanah CS UI) partner. SPLE follows the principle of mass customization, analyzing the commonalities and variabilities of the SCM system to meet diverse company needs. This approach improves the cost optimization and time efficiency in developing various SCM specifications to fulfill the requirements of each company. The development of the SCM system in this study adopts a delta-oriented programming paradigm and Abstract Behavioral Specification programming language. Subsequently, a comparison was made between the development of the SCM system using the SPLE approach and the clone-and-own approach. The research results in an enhanced SCM system developed through the SPLE, establishing it as the primary solution to existing development issues: reusing shared components and adding new custom components. Additionally, it includes an analysis that compares the SPLE approach with the clone-and-own method