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
Machine Learning Models for Predicting Flood Events Using Weather Data: An Evaluation of Logistic Regression, LightGBM, and XGBoost Maharina Maharina; Tukino Paryono; Ahmad Fauzi; Jamaludin Indra; Sihabudin Sihabudin; Muhammad Khoiruddin Harahap; Lutfi Trisandi Rizki
Journal of Applied Data Sciences Vol 6, No 1: JANUARY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i1.503

Abstract

This study examines flood prediction in Jakarta, Indonesia, a pressing concern due to its significant implications for public safety and urban management. Machine Learning (ML) presents promising methodologies for accurately forecasting floods by leveraging weather data. However, flood prediction in Jakarta remains challenging due to the city’s highly variable weather patterns, including fluctuations in rainfall, humidity, temperature, and wind characteristics. Existing methods often struggle with these complexities, as they rely on traditional ML models such as K-Nearest Neighbors (KNN), which may not capture certain patterns or provide high accuracy and robustness. Therefore, this study proposes three ML methods—Logistic Regression (LR), LightGBM, and XGBoost—to predict floods accurately. Five performance metrics (i.e., accuracy, area under the curve (AUC), precision, recall, and F1-score) were used to measure and compare the accuracy of the algorithms. The proposed method consists of three main processes. The first process involves data preprocessing and evaluation using 14 different ML models. In the second process, additional feature engineering is applied to improve the quality of the data. Finally, the third process combines the previous steps with oversampling techniques and cross-validation methods. This structured approach aims to enhance the overall performance of the analysis. The experimental results show that Process 3 significantly improves performance compared to Processes 1 and 2. The model predicts floods with an accuracy score of 93.82% for LR, 96.67% for XGBoost, and 96.81% for LightGBM, respectively. Thus, the proposed model offers a solution for operational decision-making in flood risk management, including flood mitigation planning.
Multi-Level Stacked Ensemble Learning with Multi-Search Hyperparameter Optimization for Stock Price Forecasting Imam Prasodjo; Suyanto Suyanto; Dian Indiyati; Imelda Atastina
Journal of Applied Data Sciences Vol 7, No 4: December 2026
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v7i4.1485

Abstract

Stock price forecasting remains a challenging task due to the nonlinear, dynamic, and highly volatile nature of financial markets. Traditional forecasting methods often struggle to capture complex relationships among market variables, resulting in limited predictive accuracy. This study proposes a Multi Level Stacked Ensemble Learning (MML-SEL) framework integrated with multiple hyperparameter optimization techniques to improve stock price forecasting performance. The proposed architecture combines heterogeneous machine learning models across several hierarchical stacking layers, including Random Forest, Stochastic Gradient Descent, Multilayer Perceptron, Long Short-Term Memory, eXtreme Gradient Boosting, Gradient Boosting, and Support Vector Regression. To further enhance model performance, three hyperparameter optimization methods—Grid Search, Random Search, and Bayes Search—are employed and compared. Historical stock market data consisting of Open, High, Low, Close, and Volume variables were collected from three different stock exchanges, represented by TLKM (Indonesia Stock Exchange), MSFT (NASDAQ), and BIDU (Hong Kong Stock Exchange). The dataset was chronologically divided into training and testing sets, followed by data preprocessing, feature scaling, model training, stacking integration, and performance evaluation. Forecasting accuracy was assessed using R-Squared (R²), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE). The experimental results demonstrate that the proposed MML-SEL framework achieves strong predictive performance across all datasets. The best results were obtained using Grid Search optimization, producing an R² value of 0.9996, MAE of 0.0114, and RMSE of 0.0202 on the MSFT dataset. The findings indicate that hierarchical ensemble learning combined with systematic hyperparameter optimization can significantly improve forecasting accuracy and model stability. This study contributes a scalable and effective forecasting framework that can support data-driven investment analysis and financial decision-making in diverse stock market environments.
Data Analytics of Online Lessons in Social Studies and Buddhism: Enhancing Dhamma Teaching and Tripitaka Understanding Among Teachers and Students Aammuay Luaensutthi; Thosporn Sangsawang
Journal of Applied Data Sciences Vol 4, No 3: SEPTEMBER 2023
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v4i3.125

Abstract

The objectives were to (1) determine the effectiveness of online lessons of Social Studies and Buddhism on Dhamma’s teaching regarding Tripitaka for teachers; (2) compare the pre-test and post-test achievements of teachers and primary school 6 (Grade 6) students; 3) examine the satisfaction of teachers and students using online lessons of Social Studies and Buddhism on Dharma’s teachings according to the Tripitaka. The samples were 12 teachers, and 30 students studying primary school 6 (Grade 6) at Wat Proifon School. The instruments were online lessons of the Social Studies and Buddhism course on Buddha's Teaching Tripitaka, pre-test and post-test, and the questionnaire of teachers’ and students’ satisfaction towards studying the online lessons in the Social Studies and Buddhism course on Buddha's teaching regarding the Tripitaka.Statistics used were percentage, mean, standard deviation, and t-test for dependent samples. The findings revealed that the efficiency of online lessons in the Social Studies and Buddhism course on Buddha's teaching regarding Tripitaka was 81.92/80.83 on average based on the criteria. The teachers’ learning achievements after using online lessons in the Social studies and Buddhism course on Buddha's teaching regarding the Tripitaka was higher than that of the pre-test 11.40, SD.=1.51, while the average score of the post-test was 18.17, SD.=1.10, and the t-test between   the pre-test and post-tests was 6.77, which were significantly distinctive at the level of .05., and the students’ learning achievements after using online lessons on the Social studies and Buddhism course on Buddha's teaching regarding the Tripitaka was higher than that of the pre-test: 10.40, SD.=1.61, while the average score of the post-test was 16.17, SD.=1.11, and the t-test between the pre-test and post-tests was 5.77, which were significantly distinctive at the level of .05. Teachers' satisfaction was at high level with an average of 4.47, SD.=.55, and the students’ satisfaction gained a very high level with an average of 4.50, SD.=.44.
Unsupervised Neural Networks for Breast Cancer Clustering: A Comparative Study of RBMs and SOMs with Interpretability Metrics Mekki Soundes; Labdaoui Ahlam
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.v6i4.860

Abstract

This study presents a comparative analysis of two unsupervised neural network models—Restricted Boltzmann Machines (RBMs) and Self-Organizing Maps (SOMs)—applied to breast cancer data clustering. The primary objective is to evaluate and benchmark these models in terms of their latent feature extraction, clustering accuracy, and interpretability in a medical diagnostic context. Using a preprocessed breast cancer dataset comprising 569 patient records and 30 clinical features, the models were trained and evaluated based on two internal clustering metrics: Silhouette Score and Davies-Bouldin Index (DBI). The proposed methodology, implemented in Python, emphasizes reproducibility and diagnostic relevance. RBMs achieved a Silhouette Score of 0.88 and a DBI of 0.52, indicating compact and well-separated clusters, while SOMs recorded significantly lower performance with a Silhouette Score of 0.34 and a DBI of 1.47. Furthermore, classification performance (based on cluster-label mapping) shows RBMs yielding precision between 0.82 and 0.92, and recall between 0.87 and 0.89 for benign and malignant cases. SOMs, although less accurate, offer superior visualization of high-dimensional data, which aids in exploratory analysis and interpretability. The key contribution of this work lies in the development of a standardized evaluation framework for unsupervised neural clustering in healthcare, combining quantitative clustering metrics with qualitative insights into clinical applicability. The findings demonstrate that RBMs are better suited for diagnostic tasks requiring high pattern recognition, whereas SOMs retain value for data exploration and decision explanation. This research introduces a novel integration of RBM-based clustering into medical analytics, highlighting its potential in supporting decision-making processes in oncology. Future work will extend this approach to hybrid models and multi-modal datasets, aiming to balance performance and explainability in complex diagnostic environments.
IoT based Intrusion Detection for Edge Devices using Augmented System R. Nagarajan; M. Batumalay; Zhengrui Xu
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.v5i3.358

Abstract

The Edge Computing (EC) paradigm is gaining popularity among users due to its inherent characteristics and expeditious delivery approach. Users may get information from the network's edge thanks to this feature of network architecture. The security of this edge network design, however, is a major issue. Through the Internet and in a shared setting, users can access all EC services. Intrusion detection is a method of network security that searches for threats. It is ineffective to monitor real-time network data, and current detection techniques are unable to identify known dangers. To address this problem, a technique known as augmentation oversampling is proposed, which incorporates the minority classes in the dataset. Our Sort-Augment-Combine (SAC) approach divides the dataset into subsets of the class labels, from which synthetic data is generated for each group. The developed synthetic data was then used to oversample the minority classes. After the oversampling process was complete, the distinct classes were combined to provide improved training data for model fitting. When compared to the original dataset, the models trained using the enhanced datasets perform better in terms of accuracy, recall (sensitivity), and true positives (specificity). SAC fared best in a UNSW-NB15 dataset when compared to the Synthetic Minority Oversampling Technique (SMOTE) and Generative Adversarial Network-Data Augmentation (GAN-DA). Additionally, SAC points to improvements in general sensitivity, specificity, and accuracy. SMOTE, datasets with ROSE enhancements, and Random Over-Sampling Examples for process innovation.
Psychometric Validation of an AI-Based Evaluation System for Identifying Discrepancies in Learning Processes P. Wayan Arta Suyasa; I Gusti Ngurah Pujawan; Dewa Gede Hendra Divayana; I Dewa Ayu Made Budhyani; I Made Sugiarta; I Made Candiasa
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.v7i2.1168

Abstract

This research advances the field of educational evaluation by designing and psychometrically validating an artificial intelligence (AI)- based diagnostic tool to detect discrepancies in university learning processes. The main novelty is the integration of the Provus Discrepancy Model combined with a forward-chaining inference engine. This research aims to transform evaluation from an administrative activity to an ongoing process of improvement. The tool was developed and validated through a sequential mixed-methods approach with 400 participants from 3 state universities and 8 evaluation experts. Results from the study provide evidence that the validated system created a substantial range of psychometric characteristics. These psychometric characteristics include strong content validity (SD-CVI/Ave = 0.94); high internal consistency and reliability (Cronbach's α = 0.94); solid construct validity as demonstrated through Confirmatory Factor Analysis (CFA) (CFI = 0.94; RMSEA = 0.054) and a substantial range of predictive analytics (diagnostic learning analytics), which the AI learning analytics engine evaluated learning discrepancies with a 92.4% diagnostic accuracy (47.4% more accurate than manual evaluation methods). The system's validated usefulness is demonstrated through high system usability (SUS = 88.2); high practical utility (85% total score on the Pragmatic Utility Assessment); significant utility (real-world) practical utility (detected 45 discrepancy patterns), cost efficiency (73% cost and 67% analysis time compared to traditional methods), and a range of analytics (predictive and learning discrepancy analytics). The significant contribution of this study is the development of the world's first integrated AI evaluation system that meets high methodological and psychometric standards, along with a set of real-time diagnostic analytics. Ultimately, this study developed the first truly integrated, novel paradigm evaluation system that combined the historically established evaluation construct and mechanisms with the most advanced AI capabilities, providing educators and institutions with evaluation tools to deliver data-driven pedagogical strategies and interventions in higher education. 
An Artificial Ant-Based Approach Using Polynomial Algorithms to Tackle the Text Aspect of Clustering Web Pages Souad Moufok; Khaled Belkadi; Fatima Zohra Lebbah
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.v6i2.546

Abstract

Nowadays, the web clustering problem represents a scalable research area, which is based on deep study and efficient analysis of the user's browsing behavior. Managing huge amounts of unstructured data that are given through web pages is described as a hard and primary task. In this article, we analyze clusters by grouping users based on the similarity of the web pages they have visited. Our work focuses on cleaning, analyzing, and clustering web data to facilitate users’ access to relevant content. Thus, we propose a novel algorithm, called WCLARTANT, to cluster WEB pages, which consists of finding groups of sessions according to the corresponding Web access patterns. We propose a new approach based on the ANTTREE algorithm, inspired from the self-assembling behavior observed in real ants and the binary search tree concept. The combination that we present in our approach is applied for the first time in web usage mining clustering. More precisely, different topologies are built in terms of different similarity measures, such as SBS, Euclidean, Jaccard and Cosine. Afterward, the clusters are extracted from the binary tree, which is built by the prefix depth algorithm. In other words, the proposed algorithms in this manuscript provide the corresponding binary tree to the sessions' matrix, where each node models a WEB session and each branch represents a cluster. In addition, we use the Silhouette index to evaluate and to analyze the clustering performance of WCLARTANT relative to the DBScan algorithm. WClArtAnt combined with the similarity measure SBS provides the best results compared to DBScan. The performance of our algorithm varies between 0.62 and 0.39, which are considered good. The considered log files are coming from NASA and contain all HTTP requests for a month period, from 1st July, 1995, to 31st July, 1995, for a total of 65,194 entries.
Active learning on Indonesian Twitter sentiment analysis using uncertainty sampling Muhaza Liebenlito; Nur Inayah; Esti Choerunnisa; Taufik Edy Sutanto; Suma Inna
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.v5i1.144

Abstract

Nowadays, sentiment analysis research in social media is rapidly developing. Sentiment analysis typically falls under supervised learning, which requires annotating data. However, the annotation process for sentiment analysis tasks is notoriously time-consuming. Fortunately, an effective strategy to overcome this challenge has emerged, known as active learning. Active learning involves labeling only a small subset of the dataset, leaving the rest for annotation through sampling strategies. This study focuses on comparing two active learning strategies: random sampling and boundary sampling. These strategies are applied to machine learning models such as logistic regression and random forests. In addition, we present an evaluation of the model performance and data savings achieved by implementing these strategies in the context of traditional machine learning for sentiment analysis on Twitter. The dataset considered consists of two labels: positive and negative sentiments. The results of our investigation show that active learning can significantly reduce the amount of training data required, saving up to 65% of the total training data required to achieve peak model accuracy. The most successful model identified uses a random forest with a margin sampling strategy, yielding an accuracy of 81.12% and an F1 score of 88.60%. This research highlights the effectiveness of active learning strategies in sentiment analysis, demonstrating their potential to improve model performance and resource efficiency. The results underscore the viability of employing active learning methods, particularly the combination of random forest models with margin sampling, for more efficient sentiment analysis in social media.
Brushstroke Classification from Oil Painting Images Using Convolutional Neural Networks for Tool Optimization Suraphan Chantanasut
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.v6i4.1081

Abstract

This research introduces an integrated framework that applies Convolutional Neural Networks (CNNs) to classify brushstroke types from oil painting images and utilizes the classification results to inform the design and optimization of painting tools. Researchers will conduct the brush testing activities in four sessions: Session 1: Still life painting test, Session 2: Portrait painting test, Session 3: Landscape painting test, and Session 4: Rose painting test.The classification results were mapped to specific ergonomic and functional brush design parameters, resulting in the production of ten custom-designed brush prototypes. These brushes were fabricated using precision prototyping techniques and evaluated by twenty art students and five professional artists. Quantitative user feedback revealed high satisfaction across all performance categories, including ergonomic comfort, stroke control, and paint handling. The findings confirm that CNN-based analysis of brushstroke characteristics can directly support the practical innovation of art tools, bridging computational visual analysis and traditional artistic practice. This study offers a data-driven approach to creative tool design and presents a new interdisciplinary pathway that combines deep learning, material design, and fine arts.
Biophysical Model of Mount Babaris for Predicting Carbon Potential using Remote Sensing Ahmad Jauhari; Isna Syauqiah; La Taati; Munsyi Munsyi
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.v5i4.321

Abstract

The biophysical model of Mount Babaris aims to predict carbon potential using remote sensing technology to address high levels of greenhouse gases, particularly CO2. This study combines satellite data with field measurements to create a validated model analyzing Forest Canopy Height (FCH), Normalized Difference Vegetation Index (NDVI), Vegetation Density (VD), and Land Surface Temperature (LST). A multiple regression analysis shows a strong correlation between these parameters and VD, with an R² value of 0.8673, indicating that 86.73% of the variation in vegetation density can be explained by these variables. Field validation, including drone photographs, crown and stem base density measurements, and plant size, ensures the accuracy of the satellite-derived data. The model uses the equation VD = 123.295486 x NDVI - 0.413961 x LST - 0.410253 x FCH - 3.173195, validated through field data. For processing field measurements, the equation LBDstemCor = 0.007645 x LBDcrown + 0.034348 x VD - 1.575439, with an R² value of 0.9564, further demonstrates its accuracy. To estimate carbon potential in kilograms per pixel (CPP), the equation CPP = LBDstemCor x FCHcor x 0.7 x 680 x 1.34 x 0.47 was used. The predicted carbon potential for Mount Babaris (1,576 ha) ranges from 607,767.55 to 607,829.54 tons, reflecting the model's precision in estimating carbon storage. This model plays a crucial role in monitoring and predicting carbon potential, supporting environmental management and climate change mitigation efforts. By integrating GIS and remote sensing, the model offers a scalable, replicable methodology adaptable to other regions with similar characteristics. It enhances the accuracy of carbon stock estimations and provides essential data for developing strategies to increase carbon sequestration, contributing to global climate change mitigation. The combination of satellite data, field measurements, and statistical analysis makes this model an invaluable tool for effective ecosystem conservation and restoration.