cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Rizki Wahyudi
Contact Email
rizki.key@gmail.com
Phone
+6281329125484
Journal Mail Official
telematika@amikompurwokerto.ac.id
Editorial Address
The Telematika, with registered number ISSN 2442-4528 (online) ISSN 1979-925X (print) is a scientific journal published by Universitas Amikom Purwokerto. The journal registered in the CrossRef system with Digital Object Identifier (DOI) prefix 10.35671/telematika. The aim of this journal publication is to disseminate the conceptual thoughts or ideas and research results that have been achieved in the area of Information Technology and Computer Science. Every article that goes to the editorial staff will be selected through Initial Review processes by the Editorial Board. Then, the articles will be sent to the Mitra Bebestari/ peer reviewer and will go to the next selection by Double-Blind Preview Process. After that, the articles will be returned to the authors to revise. These processes take a month for a minimum time. In each manuscript, Mitra Bebestari/ peer reviewer will be rated from the substantial and technical aspects. The final decision of articles acceptance will be made by Editors according to Reviewers comments. Mitra Bebestari/ peer reviewer that collaboration with The Telematika is the experts in the Information Technology and Computer Science area and issues around it.
Location
Kab. banyumas,
Jawa tengah
INDONESIA
Telematika
Published by Universitas Amikom Purwokerto
ISSN : 1979925X     EISSN : 24424528     DOI : 10.35671/telematika
Core Subject : Education,
Education
Jl. Letjend Pol. Soemarto No.126, Watumas, Purwanegara, Kec. Purwokerto Utara, Kabupaten Banyumas, Jawa Tengah 53127
Arjuna Subject : -
Articles 262 Documents
 23   24   25   26   27 
A Systematic Analysis of the Impact of Non-Academic Factors on Student Academic Performance Prediction Using Data Mining Gabriella Caroline Prihayu Ningsih; Febri Liantoni; Yudianto Sujana
Telematika Vol 19, No 1: February (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i1.3085

Abstract

This study investigates the prediction of students' academic performance using machine learning models through the analysis of 27 research articles. The primary objective is to identify a minimal set of essential features that significantly influence academic outcomes, aiming to optimize model performance and reduce data complexity. A Systematic Literature Review (SLR) was conducted following the PRISMA framework, focusing on key features such as midterm grades, faculty, department, demographic data, and, in some cases, behavioral attributes. The findings reveal that machine learning algorithms like Random Forest (RF) and Artificial Neural Network (ANN) consistently achieve high accuracy, surpassing 85% across various datasets, demonstrating their effectiveness in predicting academic performance. Feature selection methods, particularly filter-based techniques, were observed to significantly enhance the accuracy and efficiency of these models. Integrating diverse data, including dynamic learning behaviors, socio-economic factors, and campus attributes, is shown to further improve classification performance. Despite these advancements, challenges remain, particularly regarding the generalizability of machine learning models. Imbalanced datasets and limited dataset diversity often lead to reduced reliability when models are applied in broader contexts. Addressing these issues requires the development of more robust preprocessing techniques and advanced algorithms. The study also emphasizes the potential of deep learning models to further enhance predictive accuracy, as these approaches are capable of extracting more complex patterns from diverse datasets. Future research should prioritize expanding the scope of datasets to include a wider range of student populations and educational environments. These findings carry significant practical implications for educational institutions, enabling them to implement data-driven strategies for early intervention and personalized support. By identifying at-risk students and understanding factors influencing academic success, institutions can foster better educational outcomes and promote equitable learning opportunities.
SHAP-ALE: A Novel Approach to Explainability in Mental Health Prediction using RFR Nur Alamsyah; Budiman Budiman; Wala Erpurini; Hani Fitria Rahmani
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3087

Abstract

The prediction of mental health disorders, such as depression, has become increasingly crucial as global mental health concerns continue to rise. In this study, the predictive task specifically focuses on estimating the prevalence of depressive disorders as the primary target variable, while other mental health conditions such as anxiety and schizophrenia are treated as explanatory features. While machine learning (ML) models, like Random Forest Regressor (RFR), offer high accuracy in predictions, their interpretability remains a challenge. This research introduces SHAP-ALE, an innovative hybrid explainability framework that integrates SHapley Additive exPlanations (SHAP) and Accumulated Local Effects (ALE) to address this gap. SHAP provides both global and local insights into feature contributions, while ALE visualizes feature-target relationships, mitigating bias caused by feature correlations. Using a dataset comprising various mental health disorders and demographic factors, the dataset used in this study was obtained from Kaggle and consists of 6,421 records covering multiple mental health disorder indicators and demographic attributes across different regions and years. RFR model demonstrated robust predictive performance with an R² score of 0.9984 and a Mean Squared Error (MSE) of 0.0016. SHAP analysis revealed that features such as schizophrenia and anxiety disorders significantly influenced predictions, while ALE identified nonlinear relationships between these features and depression prevalence. The combined insights from SHAP and ALE enhance the interpretability of the model, enabling better understanding of the complex factors underlying mental health disorders. This study highlights the contribution of SHAP-ALE as a hybrid explainability framework that integrates local and global interpretability, enabling a more comprehensive understanding of feature interactions and non-linear effects beyond the capabilities of individual methods.
Enhancing Image Generation with GANs: The Role of Mutual Information in Optimizing Generative Models Sitaresmi Wahyu Handani; Pintusorn Suttiponpisarn; Gwan-yen Lin; Ruey-Feng Chang
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3304

Abstract

Generative Adversarial Networks (GANs) have become a prominent approach for image generation; however, they often suffer from training instability, mode collapse, and limited controllability of generated outputs. This study investigates the role of mutual information in improving generative modeling through a comparative analysis of Vanilla GAN, Conditional GAN (CGAN), and InfoGAN. Experiments were conducted using two image datasets with different levels of complexity, namely MNIST and Anime Face, under comparable training configurations. The evaluation focused on training behavior, convergence characteristics, generated image quality, and latent representation learning. The results revealed notable differences among the evaluated models. Vanilla GAN exhibited unstable convergence behavior at higher training epochs, while CGAN provided conditional control over generated outputs but did not fully mitigate training instability. In contrast, InfoGAN maintained more balanced generator and discriminator loss dynamics and produced visually consistent outputs across both datasets. Furthermore, latent code manipulation experiments showed that InfoGAN learned more structured and interpretable latent representations, enabling controllable feature variation in generated images. These findings indicate that incorporating mutual information improves representation learning, controllability, and training stability in GAN-based image generation. This study provides an empirical comparison of GAN, CGAN, and InfoGAN under a unified experimental framework and demonstrates that mutual information regularization contributes to improved training stability, controllable generation, and more interpretable latent representations. The findings highlight the potential of information-theoretic regularization for enhancing generative modeling performance.
Predictive Modeling for ETA and Delivery Delay Prediction in Logistics and Transportation: A Systematic Literature Review Syafrial Fachri Pane; Muhammad Qinthar Sabilla Almaliki
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3307

Abstract

This study presents a Systematic Literature Review (SLR) of data-driven predictive models for delivery delay prediction in logistics and transportation systems. A total of 508 articles were initially retrieved from the Scopus database (IEEE, Elsevier, MDPI, Springer), which served as the primary source of literature, and were systematically evaluated using the PRISMA framework through the identification, screening, eligibility, and inclusion stages, resulting in 55 selected studies published between 2016 and 2026. This review addresses two research questions: (RQ1) how data-driven predictive models, including machine learning and modern statistical approaches, are developed and applied to predict delivery delays; and (RQ2) how these models perform under varying operational conditions. The findings reveal the dominance of machine learning, deep learning, and hybrid models, which leverage heterogeneous data sources such as GPS, AIS, traffic, weather, and IoT data to capture complex spatio-temporal dependencies. Hybrid and deep learning approaches generally demonstrate superior predictive performance in dynamic and nonlinear environments, whereas conventional methods offer advantages in interpretability and computational efficiency. Model performance is strongly influenced by operational conditions, including congestion, weather variability, prediction horizon, and data quality. Inconsistent evaluation metrics limit cross-study comparability, while context-specific datasets reduce model generalizability. Dependence on historical data further constrains adaptability in real-time and disruption-prone environments. This study provides a structured synthesis of predictive modeling approaches, performance trends, and research gaps, offering guidance for researchers and practitioners in selecting and developing delivery delay prediction models. Future research should focus on integrating underutilized contextual and human-related variables, real-time multi-source data, and multi-objective optimization techniques to improve model robustness, scalability, and real-world applicability in intelligent logistics systems.
Interpretable Machine Learning for Early Detection of Academically At-Risk Students Using Behavioral, Socio-Demographic and Learning-Related Factors Vadlya Maarif; Eka Rahmawati; Candra Agustina
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3318

Abstract

The increasing adoption of learning analytics in higher education has encouraged the development of machine learning models for the early detection of academically at-risk students. However, many predictive models emphasize accuracy while providing limited interpretability for educators and academic advisors. This study proposes an interpretable machine learning framework for predicting academic risk using behavioral, socio-demographic, and learning-related student variables. A publicly available synthetic Kaggle dataset consisting of 500 student records was used as a controlled dataset for methodological validation. The data were preprocessed through missing-value handling, standardization, and one-hot encoding before being divided into training and testing sets. Several classifiers were evaluated, including Naive Bayes, Support Vector Machine, Random Forest, Logistic Regression, and XGBoost. Logistic Regression was employed as an interpretable baseline model, while XGBoost was used as a comparative ensemble classifier. Model performance was evaluated using accuracy, precision, recall, F1-score, AUC, and confusion matrix analysis. The results show that Logistic Regression achieved the highest accuracy and F1-score, with an accuracy of 0.8600 and an F1-score of 0.7812. XGBoost achieved the highest AUC value of 0.9278, followed closely by Logistic Regression with an AUC of 0.9246. Random Forest and XGBoost produced the lowest false negative values, indicating their ability to identify at-risk students more effectively. SHAP-based explainability revealed that weekly study hours, assignment completion, class attendance, and sleep duration were the most influential predictors. These findings suggest that interpretable machine learning can support academic early-warning systems by providing transparent prediction results. Since the dataset is synthetic, the findings should be interpreted as methodological validation rather than direct generalization to real student populations.
APPLICATION OF FINE-TUNED MODELS IN SENTIMENT ANALYSIS OF NEWS: A SYSTEMATIC LITERATURE REVIEW Roni Habibi; Raul Mahya Komaran
Telematika Vol 18, No 2: August (2025)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v18i2.3181

Abstract

This study aims to examine the application of fine-tuned models in news sentiment analysis through the Systematic Literature Review (SLR) approach. The main focus is directed at three aspects: improving accuracy (RQ1), implementation challenges (RQ2), and computational efficiency (RQ3). The problems identified include high computational requirements, limited annotated data, and difficulties in handling language and dialect diversity. As a solution, various optimization techniques have been explored, such as domain-specific fine-tuning, knowledge distillation, quantization, and hybrid approaches that combine fine-tuned models with lexical methods. The results of the review show that fine-tuned models, especially BERT, are capable of significantly improving sentiment analysis accuracy compared to traditional machine learning models, although they still face limitations in terms of efficiency and scalability. This study provides an important foundation for the development of more accurate, efficient, and applicable models in real-world scenarios, including news media monitoring and automated content moderation systems.
Random Forest Based Prediction of Student Stress Levels from Digital Activity Data Alphin Stephanus; Sri W. Ginting; Junus J. S. Kufla; Syukri G. Suatkab; Thenny D. Salamoni
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3416

Abstract

Early-semester students experience academic adaptation while relying intensively on digital devices, creating a need for accessible, non clinical stress screening. This study aimed to determine whether six self-reported digital-activity variables could distinguish low, moderate, and high DASS-21 derived stress categories and support a student facing screening prototype. A quantitative cross sectional survey and software prototyping design involved 66 Informatics Engineering students. The analytical procedure combined median imputation, standardization, SMOTE within each training fold, Random Forest classification, repeated stratified five fold cross validation with ten repeats, comparison algorithms, class-specific metrics, permutation importance, learning-curve analysis, and functional testing. Metric auditing produced a single out of fold accuracy of 62.12% and macro F1-score of 53.88%. Under repeated validation, the SMOTE Random Forest pipeline achieved 61.97 ± 10.92% accuracy and 55.82 ± 9.87% macro F1-score. The majority DummyClassifier obtained higher accuracy at 65.16 ± 3.47% but only 26.29 ± 0.84% macro F1-score and zero recall for moderate and high stress. SMOTE Random Forest recall was 20.67 ± 23.94% for moderate stress and 80.00 ± 30.00% for high stress. Logistic Regression produced the highest comparison-model accuracy of 70.89%, although its macro F1-score of 55.23% was slightly below that of SMOTE Random Forest. Total screen time was the only predictor with clearly positive permutation importance. Overlapping feature distributions, unstable minority class estimates, and a persistent training validation gap limited performance. StresCheck therefore constitutes an exploratory proof of concept and requires larger, balanced, externally validated data before screening deployment.
An AI Governance Framework to Address Algorithmic Bias and Educational Equity: A Systematic Literature Review Muhammad Ruslan Maulani; Saripudin Saripudin; Mumu Komaro
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3348

Abstract

The rapid integration of artificial intelligence (AI) in educational systems has generated unprecedented opportunities for personalized learning and operational efficiency. However, it has also exposed deep-rooted concerns regarding algorithmic bias and educational equity. Employing the PRISMA 2020 protocol, this systematic literature review synthesizes evidence from 46 peer-reviewed studies (2021–2026) selected from an initial pool of 102 Scopus-indexed records to examine the sources and manifestations of algorithmic bias, its differential impact on marginalized students, and the adequacy of existing AI governance frameworks in education. Findings show that bias operates through interlocking channels—biased training data, opaque model architectures, and inequitable institutional deployment—that disproportionately disadvantage students from low-income, racially minoritized, and geographically remote backgrounds, while existing governance frameworks remain largely generic and fail to address education's distinctive pedagogical and institutional dynamics. To address this gap, the study's principal contribution is the Educational AI Governance Framework (EAGF), a novel five-layer model—spanning technical accountability, institutional policy, pedagogical integration, participatory governance, and regulatory alignment—grounded in the FAIR principles (Fairness, Accountability, Inclusivity, and Responsiveness) and aligned with UNESCO, OECD, and EU AI governance standards. Unlike prior general-purpose AI ethics frameworks, the EAGF integrates these dimensions into a single coherent architecture, offering both a theoretical advance in intersectionality-aware AI governance and a practical roadmap for policymakers, educational institutions, EdTech developers, and vocational education and training (VET) systems seeking to operationalize equitable AI governance.
Cryptanalysis of Two-Factor Authentication in the Internet of Vehicles Network Environment E Haodudin Nurkifli
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3366

Abstract

An Intelligent Transportation System (ITS) plays a vital role in smart city ecosystems by enabling vehicles to communicate with roadside units, cloud servers, and other vehicles through the Internet. Numerous authentication protocols have been proposed to secure communications in the Internet of Vehicles (IoV), and recent studies have focused on enhancing user privacy through two-factor authentication. However, a detailed cryptanalysis conducted in this study reveals that an existing authentication protocol still suffers from several security vulnerabilities. Furthermore, under a stronger attacker model, an adversary who physically captures a device may extract the stored credentials, impersonate a legitimate user, and clone the device. To address these limitations, this study first presents a cryptanalysis of the existing protocol and then proposes a new authentication protocol. The proposed protocol integrates biometric authentication with a fuzzy extractor to derive cryptographic keys from fingerprint data, thereby preventing attackers from obtaining valid secret credentials even if a device is physically captured. Formal security analysis based on the Real-or-Random (RoR) model demonstrates that the proposed authentication protocol achieves strong anonymity, mutual authentication, resilience against desynchronization, and perfect forward and backward secrecy while resisting replay, impersonation, denial-of-service (DoS), and physical attacks. Furthermore, formal verification using the Scyther tool confirms that the proposed authentication protocol satisfies all specified security claims without identifying any potential attacks. Performance evaluation demonstrates that the proposed authentication protocol achieves the lowest execution time among the compared protocols, indicating its suitability for resource-constrained IoV devices.
Evaluating the Impact of SMOTE-Based Data Balancing on Decision Bias and Algorithmic Fairness in XGBoost-Based Student Dropout Prediction Les Endahti; Taqwa Hariguna; Dhanar Intan Surya Saputra
Telematika Vol 19, No 2: August (2026)
Publisher : Universitas Amikom Purwokerto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35671/telematika.v19i2.3425

Abstract

Student dropout prediction is a key application of Educational Data Mining for supporting early intervention in higher education. However, previous studies have primarily focused on improving predictive accuracy, while the effects of data balancing on decision bias and algorithmic fairness remain underexplored. This study proposes a comprehensive evaluation framework that integrates predictive performance, decision bias, and algorithmic fairness to assess the impact of the Synthetic Minority Over-sampling Technique (SMOTE) on Extreme Gradient Boosting (XGBoost) for student dropout prediction. Experiments were conducted using the publicly available Predict Students Dropout and Academic Success dataset containing 4,424 student records. After excluding the Enrolled class, the dataset was transformed into a binary classification problem consisting of 2,209 Graduate (60.9%) and 1,421 Dropout (39.1%) instances. Two models were compared: a baseline XGBoost classifier and an XGBoost classifier trained with SMOTE. Predictive performance was evaluated using Accuracy, Precision, Recall, F1-score, and ROC-AUC, while decision bias and algorithmic fairness were assessed using the False Negative Rate (FNR), Statistical Parity Difference (SPD), Disparate Impact (DI), Equal Opportunity Difference (EOD), and Average Odds Difference (AOD). The baseline model achieved higher Accuracy (93.11% vs. 92.29%), Precision (91.49% vs. 89.86%), F1-score (91.17% vs. 90.18%), and a lower FNR (0.0915 vs. 0.0951), whereas both models produced comparable ROC-AUC values (0.972). McNemar's test indicated that the difference in predictive performance was not statistically significant (p = 0.264). Although SMOTE did not improve predictive performance, it produced modest reductions in Statistical Parity Difference (0.2310–0.2218), Equal Opportunity Difference (0.0298–0.0233), and Average Odds Difference (0.0295–0.0235), indicating a slight improvement in fairness metrics while maintaining comparable discrimination capability. These findings highlight the trade-off between predictive performance and algorithmic fairness and demonstrate that evaluating predictive performance together with decision bias and fairness provides a more comprehensive assessment of educational machine learning models, supporting the development of responsible AI-based educational decision-support systems.

Page 26 of 27 | Total Record : 262

 23   24   25   26   27 

Filter by Year

2008 2026


Reset
Filter By Issues
All Issue Vol 19, No 2: August (2026) Vol 19, No 1: February (2026) Vol 18, No 2: August (2025) Vol 18, No 1: February (2025) Vol 17, No 2: August (2024) Vol 17, No 1: February (2024) Vol 16, No 2: August (2023) Vol 16, No 1: February (2023) Vol 15, No 2: August (2022) Vol 15, No 1: February (2022) Vol 14, No 2: August (2021) Vol 14, No 1: February (2021) Vol 13, No 2: Agustus (2020) Vol 13, No 1: Februari (2020) Vol 12, No 2: Agustus (2019) Vol 12, No 1: Februari (2019) Vol 12, No 1: Februari (2019) Vol 11, No 2: Agustus (2018) Vol 11, No 1: Februari (2018) Vol 10, No 2: Agustus (2017) Vol 10, No 1: Februari (2017) Vol 9, No 2: Agustus (2016) Vol 9, No 1: Februari (2016) Vol 8, No 2: Agustus (2015) Vol 8, No 1: Februari (2015) Vol 8, No 1: Februari (2015) Vol 7, No 2: Agustus (2014) Vol 7, No 1: Februari (2014) Vol 6, No 2: Agustus (2013) Vol 6, No 1: Februari (2013) Vol 6, No 1: Februari (2013) Vol 5, No 2: Agustus (2012) Vol 5, No 1: Februari (2012) Vol 4, No 2: Agustus (2011) Vol 4, No 1: Februari (2011) Vol 3, No 2: Agustus (2010) Vol 3, No 1: Februari (2010) Vol 2, No 2: Agustus (2009) Vol 2, No 1: Februari (2009) Vol 1, No 2: Agustus (2008) Vol 1, No 1: Februari (2008) More Issue