Journal of Informatics Information System Software Engineering and Applications (INISTA)

Analysis of Nginx Web Server Performance Using IPv6 with Load Balancing Method Based on Weighted Round Robin Algorithm Scheduling

Bongga Arifwidodo — Sat, 24 May 2025 00:00:00 +0800

The need for the internet affects the growth in the number of website visitors and increases the server's traffic load. The increasing number of visitors often causes the website to overload due to an excessive number of requests despite the website still using a single server. So, it is necessary to apply Load Balancing techniques. The implementation requires an algorithm, specifically the Load Balancing method, which is responsible for dividing traffic as a workload among multiple servers. This research utilizes the Weighted Round Robin (WRR) algorithm, which considers server load based on device specifications. The scenario tests optimal performance load sharing among the WRR 1:1:1, WRR 2:1:1, and WRR 3:1:1 configurations then measures Response time and CPU Utilization. Testing is performed 30 times in each test scenario, and then the average value is taken. Giving traffic loads of 1000, 2000, and 3000 Requests using H2load Benchmark. The results of the WRR 2:1:1 ratio show that it is the most optimal, as the Load is evenly distributed among the three web servers. Reading the average CPU usage for 1000-3000 Request traffic, it reaches 71%-79% on Server 1, 47%-56% on Server 2, and 48%-56% on Server 3. Then, the average Response time is 223.77ms at 1000 Requests, 233.13ms at 2000 Requests, and 235.37ms at 3000 Requests.

Data Mining Analysis of K-means Algorithm and Decision Tree for Early Detection of Students at Risk of Dropping Out

Imam Akbar, Ita Sarmita Samad, Rahmat Rahmat, Sri Rosmiana — Sat, 24 May 2025 00:00:00 +0800

Dropout occurs in higher education, where students are unable to complete their studies within a specified timeframe. It has become a significant concern in education due to its substantial impact on individuals, institutions, and society. This study aims to develop a model for predicting the early potential for students' dropout using the K-Means Algorithm and decision trees. The research method consists of a Dataset, Data Preprocessing, K-means implementation, labeling student data, and Decision Tree implementation. This study resulted in 4 clusters. The students in Cluster 1 have an excellent average GPA, a substantial number of credits, and are very active. The students in Cluster 2 have a lower average GPA and are less active than in Cluster 1. The students in Cluster 3 show a relatively good average GPA, which is lower than in Clusters 1 and 2. The number of active students indicates that students in this cluster are much less active or at risk of D.O. than those in clusters 1 and 2. Cluster 4 indicates that the average GPA of students is very low, often close to zero, and they are generally inactive in academic activities. Thus, they are significantly at risk of D.O. at Universitas Muhammadiyah Enrekang. This research provides significant results, both in terms of accuracy and data interpretation. The resulting insights enable universities to make more strategic and targeted decisions, thereby reducing the risk of university dropout rates, increasing resource efficiency, and supporting the overall educational success of students. The accuracy of the resulting model is 98.52% which indicates that the model has excellent performance in classifying students at risk of D.O.

Extracting Post‑Disaster Health Impact Information from News Reports Using Named Entity Recognition

Nalar Istiqomah, Fanny Novika — Sun, 25 May 2025 00:00:00 +0800

Natural disasters have a significant impact on public health, giving rise to various post-disaster illnesses. This study presents an automated information‑extraction framework based on Named Entity Recognition (NER), leveraging the IndoBERT model to identify disaster types, health impacts, and affected locations from online news reports. Data were gathered via web scraping from multiple reputable news portals and subsequently processed through tokenization, stop‑word removal, and lemmatization. Extracted entities were visualized via bar charts and word clouds to reveal disease patterns associated with each disaster type. Results indicate that floods have a significant public health impact, with skin diseases being the most prevalent, followed by diarrhea, fever, influenza, and Acute Respiratory Infections (ARIs). Volcanic eruptions are linked to health conditions such as ARI, hypertension, diarrhea, and influenza, whereas earthquakes show strong correlations with diarrhea, ARI, skin diseases, and fever. Droughts and landslides are closely associated with diarrheal outbreaks due to compromised sanitation resulting from limited access to clean water. Although less frequently reported, tsunamis also exhibit a notable association with cases of diarrhea. The proposed method achieves 90 % accuracy and an 88 % F1‑score. These findings confirm the effectiveness of our NER-based approach in detecting causal relationships between disasters and health outcomes, providing valuable insights for policymakers and healthcare professionals in designing targeted post-disaster mitigation and response strategies.

Classification of DDoS Attacks based on Network Traffic Patterns Using the k-Nearest Neighbor (k-NN) Algorithm

Sun, 25 May 2025 00:00:00 +0800

Many server attacks disrupt industrial or business operations. Attacks that flood bandwidth with simultaneous requests can overwhelm a system, leading to significant downtime and financial losses. Additionally, breaches that compromise sensitive data can damage a company's reputation and erode customer trust. DDoS attacks, or Distributed Denial of Service attacks, are among the most common types of server attacks. DDoS has been proven to cause server downtime, and one effective way to mitigate this attack is to detect and classify it using a machine learning approach. The K-Nearest Neighbor (KNN) algorithm, a simple yet effective classification method based on similarity measures, is known for its high accuracy. The current research builds upon two stages: the feature extraction stage and the classification stage, with the ultimate goal of improving the accuracy of DDoS identification using the CICDDoS2019 dataset. Based on this premise, the detection accuracy can be improved by enhancing these two stages. At a value of k equal to 3, this study produces an accuracy of 99.73%.

Class Balancing and Parameter Tuning of Machine Learning Models for Enhancing Aphrodisiac Herbal Plant Classification

Puguh Jayadi, Weka Sidha Bhagawan, Jofanza Denis Aldida — Sun, 25 May 2025 00:00:00 +0800

Herbal plants with aphrodisiac claims are an important part of traditional medicine that continues to evolve within the modern scientific context. However, the classification process for these plant claims is often done manually and subjectively, necessitating a more objective, data-driven approach. Artificial Intelligence (AI) and its various derivatives, such as Machine Learning, present a reliable solution for several related classification studies. The primary challenge in classification lies in data class imbalance and selecting the optimal model parameters. This study proposes an integrated approach that utilizes machine learning algorithms, including Random Forest, Support Vector Machine (SVM), and XGBoost, combined with SMOTE class balancing techniques and hyperparameter tuning through Grid Search, Random Search, and Bayesian Optimization. Experiments were conducted on a dataset of herbal plants with attributes and labels of aphrodisiac claims, and the results were evaluated based on accuracy, precision, recall, and execution time. The findings indicated that the combinatorial approach significantly improved model performance compared to the basic approach. Among the hyperparameter tuning results, the SVM method achieved the best accuracy (0.889) and precision (0.889). This research contributes to the development of an AI-based classification system in the field of ethnopharmacology. It can serve as a reference for creating scientifically validated databases of herbal plants.

Application Management Project Based on Technology Information : Study Case ASANA Evaluation of Courses Interaction Humans and Computers Surabaya State University

Azis Suroni, Bonda Sisephaputra, Saifudin Yahya — Sun, 25 May 2025 00:00:00 +0800

The development of information technology has had a significant impact on various sectors, including in the world of education, especially in project management. One of the popular information technology-based project management applications is ASANA, which offers various features to facilitate team collaboration and task management. This article aims to evaluate the use of ASANA in the Human Computer Interaction (HCI) course of the Informatics Undergraduate Study Program at Surabaya State University (UNESA) Campus 5, with a focus on ease of use, the effectiveness of team collaboration, and its impact on student work outcomes. This study uses a case study method by collecting data through questionnaires and interviews with students who use ASANA in group projects. The results of the study indicate that ASANA provides convenience in organizing tasks, speeding up communication between team members, and increasing transparency and accountability in project completion. However, several challenges such as difficulties in initial adaptation and limited features in the free version of ASANA were found in the use of this application. Overall, the ASANA application has proven effective in supporting project management in the HCI course, but there are several aspects that need to be improved to maximize its benefits. This study is expected to provide insight into the development of information technology-based project management methods in academic environments.

Application of MobileNetV2-Based Deep Learning in Detecting Diseases in Chili Plants

Wed, 02 Jul 2025 00:00:00 +0800

This study proposes a deep learning model based on MobileNetV2 architecture for the classification of chili leaf diseases using image data. The dataset was compiled from both public and private sources, covering six distinct categories of chili leaf conditions. MobileNetV2 was selected due to its efficiency and accuracy, making it ideal for real-time agricultural applications. The model was enhanced with additional layers to improve feature extraction and classification performance. Stratified 10-fold cross-validation was employed to ensure balanced evaluation across an imbalanced dataset. The experimental results showed an overall accuracy of 91.04% and an average F1-score of 0.906, indicating consistent and reliable classification performance across classes. Confusion matrix analysis highlighted strong predictive capability, particularly in detecting healthy leaves and severe disease symptoms, with minor misclassifications among visually similar categories. The findings confirm the potential of lightweight CNN architectures for practical, mobile-based agricultural diagnostics, contributing to advancements in precision farming and early disease management.

Application Decision support System for superior and high Achieving using the Analytical Hierarchy Process Method

Taruna Pratama Indarso, Cindy Taurusta, Suprianto suprianto — Mon, 26 May 2025 00:00:00 +0800

Traditional student selection methods often rely on subjective judgment, resulting in inconsistencies and potential biases. Therefore, a more systematic, objective, and efficient method is needed. The proposed system evaluates students based on four key criteria: religious competence, academic performance, extracurricular activities, and ethics. The Analytical Hierarchy Process (AHP) is used to assign priority weights to each criterion through a series of pairwise comparisons, facilitating a structured evaluation process. The Decision Support System (DSS) was developed using the Waterfall model, which includes stages of requirement analysis, system design, implementation, testing, and maintenance. Real student data from grades 7, 8, and 9 were used during system testing at Junior High School TPI Porong. This study aims to develop a mobile-based DSS to identify high-achieving students at Junior High School Taman Pendidikan Islam (TPI) Porong using the AHP method. The ranking results generated by the system were compared to manual evaluations conducted by teachers and showed over 90% consistency. Furthermore, a feasibility test involving 15 teachers indicated a 98.7% satisfaction rate, highlighting the system’s effectiveness and ease of use. The application presents rankings in a user-friendly interface, enabling teachers and school administrators to make informed decisions about student achievement. By implementing this system, schools can ensure a more transparent and data-driven process for selecting high-achieving students. The DSS not only improves the evaluation process but also supports the development of a fairer and more accountable education system. This research contributes to the advancement of technology-based educational tools that assist in decision-making within school environments.

House Sales Promotion Application Using Android-Based Augmented Reality Technology

Cahnur Saputra, Cindy Taurusta, Azmuri Wahyu Azinar — Tue, 27 May 2025 00:00:00 +0800

Augmented Reality (AR) is a technology in the field of multimedia that integrates 3D objects into real-world environments using a camera as the medium and can also be applied to mobile Android devices to enhance interactivity and visualization. This research was conducted due to shortcomings or issues in house marketing, namely the lack of detailed information about the rooms in the houses being promoted. This occurs because brochures only display the exterior of the house and are still in a 2D format. Additionally, prospective buyers who live far from the promoted housing area are unable to visit in person and cannot view the interior details of the houses being promoted or sold. Therefore, an application will be developed to visualize both the exterior and interior designs in 3D by implementing Augmented Reality technology. This is expected to make the house sales promotion for the housing area more realistic and interactive. Additionally, prospective buyers can view the exterior and interior designs of the house realistically, even without visiting the housing location directly. The house sales promotion application using Augmented Reality technology requires a camera as an input device. The application tracks and detects flat objects as markers, and after pressing the "Start" button, a 3D object that appears realistic will be automatically displayed. When the "stop" menu is pressed, the 3D image will automatically disappear. The home sales promotion application, which utilizes Augmented Reality (AR) technology, has received positive responses from respondents, achieving a high success rate. Based on the Likert scale, the application obtained an average score of 94.5%, demonstrating its effectiveness in enhancing housing promotion for potential buyers. Testing was conducted to assess the impact of AR technology in enhancing the marketing appeal and facilitating potential buyers' understanding of both the exterior and interior designs of the house more interactively and realistically.

Internet of Thing Implementation in The Library System (A Case Study of STMIK Bina Bangsa Kendari)

Faizal Aris, Sukirno Kasau — Sat, 24 May 2025 00:00:00 +0800

The subject of this article is the Internet of Things (IoT), particularly in the context of libraries. The author explains the definitions and concepts of IoT, the importance of IoT in libraries, and the potential applications of IoT for libraries. This research is located at the STMIK Bina Bangsa Kendari campus. The goal of this research is to design and simulate a prototype of an IoT-based library system utilizing UHF RFID technology. The research method used in this study is the development of a research and development model. The design of the intelligent library system development is based on two components: system hardware architecture and software development. The development approach uses a prototype development model. The library system can monitor the condition of books in real-time, whether they are available on the shelves, loaned out, or not on the shelves. The library system can provide information in the form of shelf monitoring if a book is misplaced on the shelf. The use of UHF RFID technology allows the application to read tag labels up to a maximum distance of 6 meters, while to support optimal QR Code reading in a room measuring 4 x 4 x3 meters (L x W x H), a minimum of one bulb with a power of 18 watts is required.