p-Index From 2021 - 2026
4.319
P-Index
Claim Missing Document
Check
Articles

Found 36 Documents
Search

Improving Urban Air Quality Prediction Using Bidirectional GRU: A Case Study of CO Concentration to Support Education in Yogyakarta Ordiyasa, I Wayan; Sriwidodo, Sriwidodo; Wiratma, Harits Dwi; Diqi, Mohammad; Hiswati, Marselina Endah; Noverianus, Noverianus; Syihab, Namira Anjani Rahmadina
Letters in Information Technology Education (LITE) Vol 7, No 2 (2024)
Publisher : Universitas Negeri Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.17977/um010v7i22024p63-69

Abstract

Urban air pollution, particularly carbon monoxide (CO), poses serious health risks, emphasizing the need for accurate prediction models to support real-time monitoring and timely responses. This study explores the use of a Bidirectional Gated Recurrent Unit (Bi-GRU) model to improve CO concentration forecasts, capturing intricate temporal patterns in air quality data. The model, optimized for varying input-output sequences, contributes to advancements in air quality prediction by enhancing accuracy with extended historical data. Using hourly CO data from Yogyakarta, Indonesia, the Bi-GRU model was evaluated across input lengths of 48, 96, and 144 hours with prediction outputs of 24 and 48 hours. Results show high prediction accuracy, with the best performance at 144-hour inputs, achieving an R² of 0.99 and minimal error metrics. These findings underscore the model's reliability and precision in capturing CO fluctuations, making it a promising tool for urban environmental management. This research offers a foundation for further refinement and broader applications in air quality monitoring systems.
Enhancing Medical Education with Machine Learning: A Case Study on CKD Detection Using AdaBoost-SVM Erizal, Erizal; Suwarto, Suwarto; Basuki, Umar; Hiswati, Marselina Endah; Diqi, Mohammad; Kristian, Tadem Vergi
Letters in Information Technology Education (LITE) Vol 7, No 2 (2024)
Publisher : Universitas Negeri Malang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.17977/um010v7i22024p56-62

Abstract

This study explores the application of the AdaBoost-SVM model for the classification of chronic kidney disease (CKD), addressing the critical need for accurate and early diagnosis in clinical settings. Using a dataset of 400 instances with 25 clinical features, we implemented rigorous data cleaning to remove rows with missing values, ensuring high-quality input data. The AdaBoost-SVM model achieved remarkable performance, with an overall accuracy of 96%. Precision and recall were notably high for both 'ckd' and 'notckd' classes, reflecting the model’s robustness and reliability. These results underscore the potential of hybrid machine learning approaches in medical diagnostics, providing valuable insights into improving CKD detection. Although the study has several limitations, such as a limited dataset and the exclusion of incomplete data, its findings clearly show the model's usefulness and provide a foundation for future research. Future work should focus on larger, more diverse datasets and alternative data handling techniques to enhance the model's applicability and performance. This research highlights the promise of integrating advanced algorithms into clinical decision-making processes, ultimately aiming to improve patient outcomes through early and accurate disease detection.
Optimizing Breast Cancer Detection: A Comparative Study of SVM and Naive Bayes Performance Diqi, Mohammad; Hiswati, Marselina Endah; Hamzah, Hamzah; Ordiyasa, I Wayan; Mulyani, Sri Hasta; Wijaya, Nurhadi; Wanda, Putra
TEKNOLOGI DITERAPKAN DAN JURNAL SAINS KOMPUTER Vol 7 No 1 (2024): June
Publisher : Universitas Nahdlatul Ulama Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

This study evaluates the performance of Support Vector Machine (SVM) and Naive Bayes algorithms in classifying breast cancer using the Breast Cancer Wisconsin dataset. Both models exhibited high accuracy, with Naive Bayes achieving a slightly higher overall accuracy of 97% and demonstrating a balanced performance between precision and recall. The SVM model showed strong proficiency in detecting positive cases, with an overall accuracy of 95%, though it faced minor challenges in recall for negative cases. These results highlight the effectiveness of both algorithms in breast cancer detection, emphasizing the significance of model selection based on specific diagnostic requirements. Although there are limitations, such as the small sample size and assumptions made in the model, the findings provide useful insights into the use of machine learning in medical diagnostics. This supports the idea that these models have the potential to enhance early detection and treatment results. Future research should focus on utilizing larger, more diverse datasets, exploring advanced feature processing techniques, and integrating additional algorithms to enhance further the accuracy and reliability of breast cancer detection systems.
Bootstrapped Aggregating Optimization in Random Forest for Hepatitis Risk HISWATI, MARSELINA ENDAH; DIQI, MOHAMMAD; SYAFITRI, ENDANG NURUL; FAUZIYYAH, ANNUR
Jurnal Transformatika Vol. 22 No. 1 (2024): July 2024
Publisher : Jurusan Teknologi Informasi Universitas Semarang

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26623/transformatika.v22i1.9073

Abstract

This research optimizes the Random Forest model with Bootstrapped Aggregating to predict hepatitis risk. The global significance of hepatitis as a health problem is underscored by its widespread impact. Using a Kaggle dataset comprising 596 records and 20 attributes, including age categories and gender, the study identifies limitations in predicting hepatitis risk. Through hyperparameter optimization, such as adjusting the number and depth of trees, the Random Forest model with bootstrapped aggregate achieves an accuracy of 96%, surpassing the standard model's 88%. The results demonstrate a significant improvement in precision, recall, and f1 score, particularly in reducing false negatives. The conclusion highlights the practical potential of this model for a more accurate assessment of hepatitis risk. While acknowledging limitations related to the size of the dataset, these findings provide a foundation for developing predictive models in the context of hepatitis risk, emphasizing the importance of employing ensemble techniques to improve model performance.
Optimising Bcrypt Parameters: Finding the Optimal Number of Rounds for Enhanced Security and Performance Listiawan, Indra; Zaidir, Zaidir; Winardi, Sugeng; Diqi, Mohammad
Compiler Vol 13, No 1 (2024): May
Publisher : Institut Teknologi Dirgantara Adisutjipto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.28989/compiler.v13i1.2111

Abstract

Recent advancements in the field of information security have underscored the imperative to fine-tune Bcrypt parameters, particularly focusing on the optimal number of rounds as the objective of research. The method of research is a Brute Force Search method to find the optimal value of bcrypt rounds. The primary focal point of optimization lies in the number of Bcrypt rounds due to its direct impact on security levels. Elevating the number of rounds serves to fortify the security of the Bcrypt algorithm, rendering it more resilient against brute-force attacks. The execution of the Bcrypt rounds in the experimental method mirrors real-world scenarios, specifically in the evaluation of Bcrypt parameters with a focus on entropy assessment of the hash. The selection of the number of rounds should consider the specific needs of the system, where security takes precedence or faster performance is a crucial factor.
Enhancing Diabetes Classification Using a Relaxed Online Maximum Margin Algorithm Meliala, Dyan Avando; Sulistyawati, Arum Kurnia; Diqi, Mohammad; Hiswati, Marselina Endah; Kristian, Tadem Vergi
JISKA (Jurnal Informatika Sunan Kalijaga) Vol. 10 No. 3 (2025): September 2025
Publisher : UIN Sunan Kalijaga Yogyakarta

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.14421/jiska.2025.10.3.267-278

Abstract

Diabetes mellitus is a growing global health concern that requires accurate and reliable classification models for early diagnosis and effective management. Traditional machine learning models often struggle with class imbalance, generalization limitations, and high false-positive rates, leading to misdiagnoses and delayed interventions. This study enhances the Relaxed Online Maximum Margin Algorithm (ROMMA) to improve the accuracy of diabetes classification. Using a publicly available dataset from Kaggle, which contains 768 medical records with nine health attributes, the model’s performance was evaluated through a confusion matrix and classification metrics. The Enhanced ROMMA achieved an accuracy of 92%, significantly improving upon the Standard ROMMA’s 85% accuracy. The recall for diabetes detection increased from 0.83 to 0.94, reducing false negatives and ensuring more accurate patient identification. While slight misclassification still exists, this improvement enhances the model’s reliability for clinical applications. Future research should incorporate larger datasets and advanced techniques to enhance robustness and generalizability. This study contributes to the development of more accurate machine learning models for diabetes prediction, ultimately supporting better healthcare decision-making.
Optimizing Sunspot Forecasts: An In-Depth Analysis of the ConcaveLSTM Model Ordiyasa, I Wayan; Diqi, Mohammad; Hiswati, Marselina Endah; Wandani, Aulia Fadillah Wani
International Journal of Informatics Engineering and Computing Vol. 2 No. 1 (2025): International Journal of Informatics Engineering and Computing
Publisher : ASTEEC

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.70687/ijimatic.v2i1.103

Abstract

This work examines how effectively the ConcaveLSTM model can forecast sunspot numbers, recognizing their importance in space weather. The model addresses the complex and changing sunspot characteristics to improve forecasting accuracy. By comparing different model variations, this research identifies optimal combinations of input steps and LSTM units that enhance forecast performance while avoiding overfitting. The study showcases the capability of specific architectures concerning detail versus computational cost, using evaluation metrics such as RMSE, MAE, MAPE, and R2. Considering factors like limited data availability and the complexity of solar phenomena, the ConcaveLSTM model could be a valuable tool for predicting solar activity. This research advances understanding of space weather forecasting through machine learning and offers guidance for further model development and future investigations.
PROTEGO: Improving Breast Cancer Diagnosis with Prototype-Contrastive Autoencoder and Conformal Prediction on the WDBC Dataset Hiswati, Marselina Endah; Diqi, Mohammad
Jurnal Teknik Informatika (Jutif) Vol. 6 No. 5 (2025): JUTIF Volume 6, Number 5, Oktober 2025
Publisher : Informatika, Universitas Jenderal Soedirman

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52436/1.jutif.2025.6.5.5294

Abstract

Breast cancer remains one of the leading causes of mortality among women, making accurate and trustworthy early detection a critical challenge in healthcare. To address this, we propose PROTEGO, a Prototype-Contrastive Autoencoder with integrated Conformal Prediction, designed to achieve both high diagnostic accuracy and reliable uncertainty quantification. The framework combines dual-head autoencoding, supervised contrastive learning, prototype-based regularization, and conformal calibration to generate discriminative yet interpretable representations. Using the Wisconsin Diagnostic Breast Cancer (WDBC) dataset, PROTEGO was trained and evaluated through stratified data splits, with performance measured by AUROC, AUPRC, F1-score, Balanced Accuracy, Brier score, calibration error, and conformal coverage metrics. The results show that PROTEGO achieves highly competitive performance with an AUROC of 0.992 and an AUPRC of 0.995, while uniquely providing conformal coverage guarantees with an average set size close to one and more than 92% decisive predictions. Ablation studies confirm the complementary role of each component in enhancing both accuracy and calibration. These findings demonstrate that integrating prototype-guided representation learning with conformal prediction establishes a clinically meaningful diagnostic framework. PROTEGO highlights the importance of unifying precision and reliability in medical AI, offering a step toward more interpretable, safe, and clinically trustworthy systems for breast cancer detection.
Stacked Gated Recurrent Units and Indonesian Stock Predictions: A New Approach to Financial Forecasting DIQI, MOHAMMAD; HISWATI, MARSELINA ENDAH; WIJAYA, NURHADI
Jurnal IT UHB Vol 5 No 1 (2024): Jurnal Ilmu Komputer dan Teknologi
Publisher : Universitas Harapan Bangsa

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.35960/ikomti.v5i1.1106

Abstract

This research paper introduces a novel approach to predicting stock prices using a Stacked Gated Recurrent Unit (GRU) model. The model was trained on historical data from the top 10 companies listed on the Indonesia Stock Exchange, covering the period from July 6, 2015, to October 14, 2021. The performance of the model was evaluated using key metrics, including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and R-squared (R2). The results demonstrated promising performance, with average RMSE, MAE, and MAPE values of 0.00592, 0.00529, and 0.01654, respectively, indicating a high level of accuracy in the model's predictions. The average R2 value of 0.97808 further suggests a high degree of predictive power, with the model able to explain a significant proportion of the variance in the stock prices. These findings highlight the effectiveness of the Stacked GRU model in capturing stock price patterns and making accurate predictions. The practical implications of this research are significant, as the model provides a powerful tool for forecasting future stock price trends, which can be utilized in investment decision-making, financial analysis, and risk management. Future research could explore other deep learning architectures, incorporate additional features, or consider different evaluation metrics to enhance the model's performance further.
Multi-Step Vector Output Prediction of Time Series Using EMA LSTM Diqi, Mohammad; Sahal, Ahmad; Nur Aini, Farida
JOIN (Jurnal Online Informatika) Vol 8 No 1 (2023)
Publisher : Department of Informatics, UIN Sunan Gunung Djati Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.15575/join.v8i1.1037

Abstract

This research paper proposes a novel method, Exponential Moving Average Long Short-Term Memory (EMA LSTM), for multi-step vector output prediction of time series data using deep learning. The method combines the LSTM with the exponential moving average (EMA) technique to reduce noise in the data and improve the accuracy of prediction. The research compares the performance of EMA LSTM to other commonly used deep learning models, including LSTM, GRU, RNN, and CNN, and evaluates the results using statistical tests. The dataset used in this study contains daily stock market prices for several years, with inputs of 60, 90, and 120 previous days, and predictions for the next 20 and 30 days. The results show that the EMA LSTM method outperforms other models in terms of accuracy, with lower RMSE and MAPE values. This study has important implications for real-world applications, such as stock market forecasting and climate prediction, and highlights the importance of careful preprocessing of the data to improve the performance of deep learning models.