The selection of structural officials within higher education institutions is a strategic and complex process that demands objectivity, transparency, and a data-driven approach. However, the increasing number of candidates and the diversity of evaluation criteria, such as years of service, rank, education, age, and performance, pose significant challenges in ensuring fair and efficient decision-making. Addressing this gap, this study proposes a hybrid method by integrating Skyline Query with the Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE), offering a novel contribution to multi-criteria decision-making (MCDM) in public sector human resource selection. Skyline Query is employed as a preselection mechanism to eliminate 161 dominated candidates from an initial dataset of 228, allowing only the 67 most non-dominated candidates to advance to the ranking stage. PROMETHEE is then applied to generate rankings based on leaving and entering flow values. To evaluate the consistency and validity of this combined approach, the resulting rankings are compared with those from the pure PROMETHEE method using Spearman’s Rank Correlation. The analysis yields a high correlation coefficient of ρ = 0.967, indicating a very strong agreement between the two methods and confirming that the Skyline filtering does not distort ranking quality. The findings demonstrate that the Skyline+PROMETHEE integration significantly enhances the efficiency of the selection process by reducing computational complexity while preserving decision accuracy. Moreover, this approach strengthens the transparency and accountability of structural official selection, particularly in the context of the University of Mataram, and can be generalized to other institutional decision-making scenarios.

Climate change is a global long-term change in temperatures and weather. Climate change is a worldwide issue that requires proper handling to reduce the negative impact on humans and the environment. Analyzing historical data is beneficial for studying climate change. Machine learning and deep learning methods are useful tools for data analysis. The goal of this paper is to find the best model for forecasting temperatures, a case study in Java Island. Java Island is the most densely island and the central economy and business in Indonesia. Climate change research in Java Island is important for sustainability. It runs several algorithms i.e., Gradient Boosting, AdaBoost, XGBoost, CatBoost, Light GBM, Random Forest, Support Vector Regression, Extreme Learning Machine, Long Short-Term Memory, Gated Recurrent Unit, Bidirectional Long Short-Term Memory, and Bidirectional Gated Recurrent Unit. The experiment uses a historical daily time series of temperatures from 1 January 1990 to 31 December 2024. In general, the experimental results show that Gradient Boosting produces the highest average coefficient of determination R2 scores of 0.34 and the lowest Mean Absolute Error scores of 0.69. Long Short-Term Memory and Gated Recurrent Units are the deep learning models that also work well for forecasting. According to the experimental results, in some cases, machine learning models outperform deep learning models and vice versa.

Rice farming productivity worldwide has been declining due to improper soil management practices, including excessive chemical fertilizer use and irregular irrigation. The main challenge lies in accurately classifying soil fertility levels to support optimal land use and reduce resource waste, especially when dealing with imbalanced datasets. This study aims to compare the performance of single classifiers and ensemble classifiers in classifying soil fertility. The single classifiers used include K-Nearest Neighbor (KNN), Naive Bayes, Decision Tree, Support Vector Machine (SVM), and Artificial Neural Network (ANN), while the ensemble classifiers include Random Forest and XGBoost. The Indian Soil Fertility Dataset, obtained from Kaggle, contains 880 samples with 12 features and 1 output class. The research methodology involved data acquisition, preprocessing, data splitting, standardization, and classification, with performance evaluation conducted using a confusion matrix. The results show that ensemble classifiers, particularly Random Forest and XGBoost, outperform single classifiers in imbalanced datasets, achieving accuracy, precision, recall, and F1-score values exceeding 92%-95% across all split scenarios. The findings conclude that Random Forest and XGBoost can serve as reliable models for assisting farmers and agricultural experts in evaluating soil conditions, minimizing unnecessary fertilizer usage, and improving rice farming productivity globally.

The issue of waste management remains a critical concern due to its adverse impact on the environment. This research enhances a deep learning-based waste classification model by introducing a new class, namely plastic sachets, to broaden the classification scope and increase the model's relevance to waste types commonly found in the community. The dataset used is an extended version of a previous open-source dataset, comprising 2,968 images divided into seven classes. Data preprocessing steps include stratified data splitting, data augmentation to increase image diversity, and pixel normalization. The model adopts the MobileNetV2 architecture through a transfer learning approach, utilizing 2D Global Average Pooling and Dense layers with softmax activation for multi-class classification. Evaluation using precision, recall, and F1-score demonstrated strong performance, with an overall accuracy of 97%. While the model performs well across most classes, further improvement is needed for minority classes such as plastic sachets. This study highlights the promising potential of deep learning in supporting automated waste sorting to promote sustainable waste management practices in Indonesia.

Conventional physics laboratories often suffer from limitations in terms of equipment availability and safety, which hinders optimal learning of collision dynamics concepts. This research aims to develop a virtual laboratory based on collision algorithm to simulate perfect collision as an alternative solution in physics learning. The development uses the ADDIE model, which includes the stages of analysis, design, development, implementation, and evaluation. The collision algorithm was implemented using ActionScript 3, with interpolation allowing for more accurate collision detection at high speeds. The validation results show that the simulation is in line with the law of conservation of momentum and kinetic energy and is consistent with analytical solutions from MATLAB and Python. Functionality testing was conducted by 20 students, and the results showed that the use of this virtual laboratory significantly improved their concept understanding, with the average improvement ranging from 24% to 56%. Students also reported that this virtual laboratory is more interactive and interesting, thus increasing their learning motivation. The conclusion of this study is that the collision algorithm-based virtual laboratory is effective as a physics learning media and can be adopted more widely in technology-based education, especially to understand complex physics concepts more deeply.

The need for structured and semantically rich data in higher education underscores the role of ontology-based knowledge modeling. This study develops an ontology to represent national-level student achievements, covering key aspects such as institution, achievement field, category, year, level, and student status. Using a formal ontology engineering approach, the ontology was developed in Protégé and encoded in OWL. Evaluation involved technical validation and reasoning tests including class subsumption, consistency checking, instance classification, and rule-based inference to assess logical soundness and semantic correctness. Description Logic (DL) queries were also executed based on competency questions to evaluate the ontology’s ability to support semantic querying. The results demonstrate that the ontology effectively supports knowledge inference and structured data retrieval, offering strong potential for integration within semantic web environments. This provides a foundation for data interoperability and knowledge sharing across educational systems at the national level. Future work includes expanding the ontology to incorporate dynamic achievement updates and linking with external educational data sources.

A Supply Chain Management (SCM) application prototype for PT XYZ has been developed. Before its release, the testing phase is crucial to minimize functional failures. This study aims to conduct comprehensive testing to verify that all functions operate as intended, deliver the expected results, and identify potential errors in the application. Unlike previous research, this study not only uses Equivalence Partitioning as a functional testing technique but also combines it with the ISO/IEC 25010 standard, which guides Functional Suitability and software quality more comprehensively. This approach addresses the limitations of Equivalence Partitioning, which typically focuses only on input class division without ensuring coverage of specific quality standards. Furthermore, integrating testing techniques with the internationally recognized ISO/IEC 25010 standard enhances the relevance and applicability of this research in software development for industries, particularly in supply chain information processing automation. By combining both methods, the testing process becomes more thorough and structured, as evidenced by the 254 test cases generated. Additionally, this study provides quantitative data in the form of valid (83%) and invalid (17%) test case percentages, which can serve as indicators of testing effectiveness and the prototype's quality.

Multiple object detection remains a significant challenge in the field of computer vision. One of the key factors affecting detection performance is the feature extraction process, especially when objects are relatively small or positioned closely together. This study aims to compare the effectiveness of two popular object detection models, YOLO (You Only Look Once) and Single Shot MultiBox Detector (SSD), in detecting multiple objects within images. These models were selected due to their reported high accuracy and real-time processing capabilities, outperforming traditional methods such as the Hough Transform, Deformable Part-based Models (DPM), and conventional CNN architectures. The models were evaluated using a subset of the PASCAL VOC dataset, which includes object categories such as aircraft, faces, cars, and others, with a total of 1,447 annotated images used in training and testing. The evaluation metric used was mean Average Precision (mAP) to assess detection accuracy. Experimental results indicate that YOLO achieves a mAP of 82.01%, while SSD achieves 70.47%. These findings demonstrate that YOLO provides better performance in detecting multiple objects under the same conditions. Overall, this study confirms the advantages of YOLO in scenarios requiring fast and accurate multi-object detection, highlighting its potential for deployment in real-time applications such as autonomous vehicles, surveillance systems, and robotics. The main contribution of this study lies in providing a comparative performance benchmark between YOLO and SSD on a standard multi-object dataset to guide practical model selection in real-time computer vision tasks.