cover
Contact Name
Husni Teja Sukmana
Contact Email
husni@bright-journal.org
Phone
+62895422720524
Journal Mail Official
jads@bright-journal.org
Editorial Address
Gedung FST UIN Jakarta, Jl. Lkr. Kampus UIN, Cemp. Putih, Kec. Ciputat Tim., Kota Tangerang Selatan, Banten 15412
Location
Kota adm. jakarta pusat,
Dki jakarta
INDONESIA
Journal of Applied Data Sciences
Published by Bright Publisher
ISSN : -     EISSN : 27236471     DOI : doi.org/10.47738/jads
One of the current hot topics in science is data: how can datasets be used in scientific and scholarly research in a more reliable, citable and accountable way? Data is of paramount importance to scientific progress, yet most research data remains private. Enhancing the transparency of the processes applied to collect, treat and analyze data will help to render scientific research results reproducible and thus more accountable. The datasets itself should also be accessible to other researchers, so that research publications, dataset descriptions, and the actual datasets can be linked. The journal Data provides a forum to publish methodical papers on processes applied to data collection, treatment and analysis, as well as for data descriptors publishing descriptions of a linked dataset.
Articles 1,270 Documents
Clustering-Based Adaptive UX in E-Learning Systems: Aligning Microservices with the 4C Framework Poetri Lestari Lokapitasari Belluano; Syaad Patmanthara; Muhammad Ashar; Fachrul Kurniawan; Gulsun Kurubacak
Journal of Applied Data Sciences Vol 6, No 4: December 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i4.884

Abstract

This study introduces a clustering-driven adaptive User Experience (UX) architecture for e-learning systems, aligning machine learning segmentation with the 21st-century 4C educational framework (critical thinking, communication, collaboration, creativity). The objective is to dynamically personalize digital learning interactions through a microservices architecture responsive to users' UX profiles. A quantitative survey was conducted involving 50 active users of Shopee and Tokopedia, whose interaction feedback was mapped using the User Experience Questionnaire (UEQ). Three unsupervised clustering techniques—KMeans, Agglomerative, and DBSCAN—were compared. KMeans outperformed the others with a silhouette score of 0.157, compared to 0.146 for Agglomerative and −0.017 for DBSCAN, identifying three meaningful clusters representing high, medium, and low UX proficiency. A one-way ANOVA test confirmed statistically significant differences (p 0.01) among the clusters in dimensions such as error clarity, support responsiveness, and user confidence. These UX profiles were then mapped to individualized microservices: Cluster 0 received autonomous content with minimal support, Cluster 1 was offered guided prompts, and Cluster 2 was provided with simplified interfaces and proactive assistance. Each cluster was aligned with specific 4C competencies to ensure pedagogical relevance. The proposed architecture, built with gRPC-based microservices, enabled asynchronous, low-latency personalization based on user cluster membership. The novelty of this research lies in its dual alignment—technological (microservices + machine learning) and educational (4C competency mapping)—to construct a scalable and responsive e-learning environment. The system design, although validated through simulation, demonstrates a practical foundation for future deployment in platforms like Moodle or OpenEdX. By linking behavioral UX clustering to pedagogical intervention strategies, this study offers a model for adaptive, data-informed instructional systems that are both scalable and learner-centered.
Unveiling Criminal Activity: a Social Media Mining Approach to Crime Prediction Sheeba Armoogum; Deshinta Arrova Dewi; Vinaye Armoogum; Nicolas Melanie; Tri Basuki Kurniawan
Journal of Applied Data Sciences Vol 5, No 3: SEPTEMBER 2024
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v5i3.350

Abstract

Social media platforms have become breeding grounds for abusive comments, necessitating the use of machine learning to detect harmful content. This study aims to predict abusive comments within a Mauritian context, focusing specifically on comments written in Mauritian Kreol, a language with limited natural language processing tools. The objective was to build and evaluate four machine learning models—Decision Tree, Random Forest, Naïve Bayes, and Support Vector Machine (SVM)—to accurately classify comments as abusive or non-abusive. The models were trained and tested using k-fold cross-validation, and the Decision Tree model outperformed others with 100% precision and recall, while Random Forest followed with 99% accuracy. Naïve Bayes and SVM, although achieving 100% precision, had lower recall rates of 35% and 16%, respectively, due to imbalanced data in the training set. Pre-processing steps, including stop-word removal and a custom Kreol spell checker, were key in enhancing model performance. The study provides a novel contribution by applying machine learning in a Mauritian context, demonstrating the potential of AI in detecting abusive language in underrepresented languages. Despite limitations such as the absence of a Kreol lemmatization tool and incomplete coverage of Kreol spelling variations, the models show promise for wider application in social media crime detection. Future research could explore expanding this approach to other languages and domains of social media crimes.
A Hybrid Multi-Criteria and Factorial Analysis Framework for CIRFLINK CubeSats Communication Design Warinthorn Kiadtikornthaweyot Evans; Ratchanon Chuphet
Journal of Applied Data Sciences Vol 7, No 2: May 2026
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v7i2.1101

Abstract

This research presents the design and optimization of a low-cost Inter-Satellite Link for the CIRFLINK project, a 1.5U CubeSat mission aimed at advancing smart farming and disaster monitoring in rural area. To address complex design trade-offs, this study introduces a novel integrated framework combining the Analytic Hierarchy Process for qualitative prioritization and 2k factorial design for quantitative validation. The primary objective was to evaluate and optimize LoRa-based communication parameters for constrained satellite environments. Analytic Hierarchy Process results prioritized LoRa technology over optical and traditional RF subtypes due to its superior power efficiency and simplicity. Subsequently, 2k factorial experiments and ANOVA revealed that distance and physical obstruction are the dominant factors affecting performance, while parameters like Spreading Factor and Bandwidth showed less immediate impact in the tested ranges. Experimental results using ESP32 and SX1278 modules demonstrated that the Signal-to-Noise Ratio and Received Signal Strength Indicator maintain reliable connectivity, with correlation analysis showing a strong negative relationship approximate -0.913b etween distance and signal quality. Field data confirmed that the system achieves stable communication with an average SNR in line-of-sight conditions. The novelty of this work lies in the systematic fusion of multi-criteria decision-making with statistical experimental design, providing a transparent engineering roadmap for small-satellite communication. This contribution offers a validated, cost-effective ISL solution that meets mission requirements with minimal complexity, serving as a scalable model for future educational and IoT-based CubeSat constellations.
Detection of COVID-19 using EfficientnetV2-XL and Radam Optimizer from Chest X-ray Images Ibrahim Alkore Alshalabi; Tawfiq Alrawashdeh; Sumaya Abusaleh; Malek Zakarya Alksasbeh; Khalid Alemerien; Shorouq Al-Eidi; Hamzah Alshamaseen
Journal of Applied Data Sciences Vol 6, No 2: MAY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i2.512

Abstract

Automating the detection of the COVID-19 pandemic has become necessary for assisting radiologists and medical practitioners in the diagnosis process. It enables them not only to save time through early diagnosis but also to ensure that they are making more accurate diagnoses. Therefore, this research presents a novel approach for automatically identifying COVID-19 in chest X-ray images by utilizing the EfficientNetV2-XL model in combination with the Rectified Adam optimizer for training. For conducting the experiments, we used the dataset available on Kaggle, known as the “COVID-19 Radiography Dataset.” The totality of this dataset was 21,165, and it included four patterns: COVID-19, viral pneumonia, lung opacity, and normal cases. The dataset was divided into 80% training and 20% testing. The preprocessing stage included resizing images to 512 × 512 pixels and then applying data augmentation techniques to enhance model robustness. Consequently, a fine-tuned multiclass categorization system was implemented. The proposed system's effectiveness is evidenced by the experimental outcomes, which show a 99.31% accuracy rate and a perfect Area Under the Curve score of 1 for identifying COVID-19. Additionally, the Score-CAM visualization method was utilized to enhance the interpretability of model predictions, identifying key regions within the chest X-ray images that influence the classification outcome. This Localization technique aids healthcare professionals in understanding the reasoning behind the model and confirming the accuracy of the diagnosis. The proposed system outperformed the state-of-the-art models for COVID-19 detection. 
Exploring ADR Trends: A Data Mining Approach to Hotel Room Pricing, Cancellations, and EDA Nina Kurnia Hikmawati; Yudi Ramdhani; Wartika Wartika
Journal of Applied Data Sciences Vol 5, No 1: JANUARY 2024
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v5i1.165

Abstract

This study investigates the intricacies of hotel reservation cancellations by analyzing a comprehensive dataset that includes information from both City Hotel and Resort Hotel. Through a thorough examination of various aspects, the research provides detailed insights into cancellation tendencies, daily rates, seasonal trends, and the influence of geographic factors and market segments on cancellation behavior. The overall cancellation and non-cancellation ratios indicate a notable non-cancellation rate of 62.86%, showcasing a high level of guest confidence in their reservations. Conversely, the 37.14% cancellation ratio raises concerns about potential negative repercussions. A comparative analysis between City Hotel and Resort Hotel reveals a significant difference in cancellation rates, emphasizing the need for tailored strategies at City Hotel to enhance booking stability. The study on Average Daily Rate (ADR) for both hotels bring attention to price differences and seasonal trends. Resort Hotel's higher ADR suggests potential advantages in location or amenities. Seasonal trends, particularly the highest ADR during the summer, provide valuable insights for resource planning. The variation in cancellation rates based on countries emphasizes the importance of focused strategies in regions with high cancellation rates, as seen with Portugal having the highest cancellation rate (77.70%). Analysis of hotel customer market segments identifies Online Travel Agencies (OTA) as the segment with the highest cancellation rate (46.97%). These findings present opportunities for tailored marketing and cancellation policies based on the characteristics of each segment. In conclusion, this research offers strategic insights for hotel managers to enhance booking stability, design competitive pricing policies, and understand the impact of geographic factors and market segments on cancellation behavior.
Classification of Batak Toba Ulos Motifs Based on Transfer Learning with MobileNetV2 Tonni Limbong; Gonti Simanullang; Parasian DP. Silitonga; Donalson Silalahi
Journal of Applied Data Sciences Vol 6, No 4: December 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i4.1036

Abstract

Indonesia possesses a rich cultural heritage, including the traditional Batak Toba Ulos textile, which is known for its diverse motifs and deep philosophical meanings. However, the preservation and visual recognition of Ulos remain challenging, particularly in terms of systematic documentation and automated classification. This study presents a visual recognition system for Batak Toba Ulos motifs using a transfer learning approach based on the MobileNetV2 architecture. The methodology involves the construction of a curated dataset of Ulos images, the application of data augmentation and preprocessing techniques, and model training utilizing ImageNet pre-trained weights. The system’s performance was evaluated using accuracy, precision, recall, and F1-score metrics. Results show that the model is capable of accurately classifying all 12 Ulos classes, achieving F1-scores ranging from 0.93 to 0.97. These findings demonstrate that transfer learning is effective in overcoming the limitations of culturally specific, small-scale datasets. This research contributes to the development of artificial intelligence tools for cultural preservation and supports the digital documentation and promotion of Batak Toba Ulos to younger generations and broader audiences in an efficient and scalable manner.
How Effective are Different Machine Learning Algorithms in Predicting Legal Outcomes in South Africa? Joe Khosa; Daniel Mashao; Ayorinde Olanipekun; Charis Harley
Journal of Applied Data Sciences Vol 5, No 4: DECEMBER 2024
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v5i4.215

Abstract

This study examines the effectiveness of different machine learning algorithms in predicting legal outcomes in South Africa's Judiciary system. Considering the advancement of artificial intelligence in the legal sector, this research aims to assess the effectiveness of various machine learning algorithms within the legal domain. Text classification is done using machine learning algorithms, including Logistic Regression, Random Forest, and K-Nearest Neighbours, with datasets obtained from a state legal firm in South Africa. The datasets undergo diligent data cleansing and pre-processing methods, encompassing tokenization and lemmatization techniques. This study evaluates these models' applications through accuracy metrics. The findings demonstrate that the Logistic Regression model attained an accuracy rate of 75.05%, whereas the Random Forest algorithm achieved an accuracy rate of 75.08%. On the other hand, the K-Nearest Neighbours algorithm exhibited no optimal performance, as evidenced by its accuracy rate of 62.76%. This study provides valuable insights for legal professionals by addressing a specific research question about the successful application of machine learning in South Africa's legal sector. The results indicate the possibility of using machine learning to predict the outcomes of criminal legal cases. Additionally, this study highlights the significance of responsibly and ethically implementing machine learning within the legal field. The results of this study enhance our comprehension of the prediction of legal outcomes, establishing a foundation for future investigations in this dynamic area of study. A limitation of this study is that the data was obtained from a single law firm in South Africa.
FCI-ANTREE: A GUI-Centric Method for Schema Recovery and Conceptual Database Model Reconstruction in Legacy Form-Based Systems Juanda Hakim Lubis; Elviawaty Muisa Zamzami; Mahyuddin K. M Nasution; Mohammad Andri Budiman
Journal of Applied Data Sciences Vol 7, No 3: September 2026
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v7i3.1383

Abstract

Legacy systems that lack technical documentation present significant challenges for database schema recovery, particularly when access to source code and SQL queries is unavailable. Existing reverse engineering approaches predominantly rely on backend artifacts such as database logs, schema definitions, or program code, limiting their applicability in undocumented environments. Although GUI-based approaches offer an alternative by utilizing interface-level information, many existing methods still rely on shallow visual parsing and lack systematic mechanisms to capture structural and interaction semantics. To address this limitation, this study proposes FCI–ANTREE, a GUI-centric method for reconstructing conceptual database schemas from legacy form-based systems. The method integrates Form-Centric Interaction (FCI) to extract candidate entities, attributes, relationships, constraints, and data types from user interactions and validation logic, and Admin Interface Tree (ANTREE) to model hierarchical relationships and transform them into logical and relational schemas. The objective of this research is to provide a systematic, interpretable, and semi-automated approach for database reverse engineering without relying on backend access. The proposed method was evaluated using three case studies: an online store application, a library system, and an inventory application. The evaluation employed structural consistency analysis, confusion-matrix-based metrics, and quantitative error measurements. The results show that the method achieved a mean MAE of 1.77, RMSE of 3.16, and R² of 97.09%, along with an average F1-score of 0.8768, indicating a high level of agreement between reconstructed and reference schemas. These findings demonstrate that FCI–ANTREE provides an effective and practical solution for database schema reconstruction in legacy systems with limited or no backend accessibility. The method contributes by introducing an interaction-aware and rule-based framework that enhances the accuracy, interpretability, and applicability of GUI-driven reverse engineering.
Comparing Epsilon Greedy and Thompson Sampling model for Multi-Armed Bandit algorithm on Marketing Dataset Izzatul Umami; Lailia Rahmawati
Journal of Applied Data Sciences Vol 2, No 2: MAY 2021
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v2i2.28

Abstract

A/B checking is a regular measure in many marketing procedures for e-Commerce companies. Through well-designed A/B research, advertisers can gain insight about when and how marketing efforts can be maximized and active promotions driven. Whilst many algorithms for the problem are theoretically well developed, empirical confirmation is typically restricted. In practical terms, standard A/B experimentation makes less money relative to more advanced machine learning methods. This paper presents a thorough empirical study of the most popular multi-strategy algorithms. Three important observations can be made from our results. First, simple heuristics such as Epsilon Greedy and Thompson Sampling outperform theoretically sound algorithms in most settings by a significant margin. In this report, the state of A/B testing is addressed, some typical A/B learning algorithms (Multi-Arms Bandits) used to optimize A/B testing are described and comparable. We found that Epsilon Greedy, be an exceptional winner to optimize payouts in this situation.
Assimilate Grid Search and ANOVA Algorithms into KNN to Enhance Network Intrusion Detection Systems Mohammad A. Alsharaiah; Mohammed Amin Almaiah; Rami Shehab; Tayseer Alkhdour; Rommel AlAli; Fares Alsmadi
Journal of Applied Data Sciences Vol 6, No 3: September 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i3.604

Abstract

The recent progress of operational network intrusion detection systems (NIDS) has become increasingly essential. Herein, a fruitful attempt to introduce an innovative NIDS methodology that integrates the grid search optimization algorithm and ANOVA techniques with the K nearest neighbor (KNN) algorithm to analyze both spatial and temporal characteristics of data for network traffic. We employ the UNSW-NB15 benchmark dataset, which presents various patterns and a notable imbalance between the training and testing data, with 257674 samples. Therefore, the Synthetic Minority Oversampling Technique has been used since this method is effective in handling imbalanced datasets. Further, to handle the overfitting issue the K folds cross-validation method has been applied. The feature sets within the dataset are meticulously selected using ANOVA mechanisms. Subsequently, the KNN classifier is fine-tuned through hyperparameter tuning using the grid search algorithm. This tuning process includes adjusting the number of K neighbors and evaluating various distance metrics such as 'euclidean', 'manhattan', and 'minkowski'. Herein, all attack types in the dataset were labeled as either 1 for abnormal instances or 0 for normal instances. Our model excels in binary classification by harnessing the strengths of these integrated techniques. By conducting extensive experiments and benchmarking against cutting-edge machine learning and deep learning models, the effectiveness and advantages of our proposed approach are thoroughly demonstrated. Achieving an impressive performance of 99.1%. Also, several performance metrics have been applied to assess the proposed model's efficiency.