Claim Missing Document
Check
Articles

Found 3 Documents
Search

COMPARISON OF MACHINE LEARNING CLUSTERING ALGORITHMS FOR ANALYSING ELECTRICITY USAGE PATTERNS IN CAMPUS AREAS Purba, Diya Namira; Muhammad Ridha; Rida Indah Fariani; Harkiapri Yanto
Jurnal Teknoif Teknik Informatika Institut Teknologi Padang Vol 13 No 2 (2025): TEKNOIF OKTOBER 2025
Publisher : ITP Press

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21063/jtif.2025.V13.2.87-96

Abstract

Electricity consumption in campus environments varies based on building functions, occupancy patterns, and time-of-day usage. Understanding these variations is essential for efficient energy management. Uncontrolled electricity use often results in high operational costs, highlighting the need for accurate methods to uncover consumption patterns. This study analyzes electricity consumption data from multiple campus buildings at a polytechnic in Jakarta during 2023 and 2024. Each dataset consists of six columns and 365 rows in a year. Since the data is unlabeled, three clustering algorithms: K-Means, Hierarchical Clustering, and DBSCAN are applied to identify usage patterns across campus areas. Pre-processing included imputation and normalization, followed by clustering. Cluster quality was evaluated using the Silhouette Score. A key novelty of this study is the year-to-year comparative analysis, showing that clustering performance can vary significantly depending on data structure and noise. The 2023 dataset (dataset 1) achieved the highest Silhouette Score of 0.48 using DBSCAN, while the 2024 dataset (dataset 2) produced the best result with Hierarchical Clustering at 0.53. These results emphasize the importance of selecting clustering methods based on data characteristics and temporal context. The findings contribute to developing adaptive, data-driven strategies for managing energy use in non-residential settings, particularly in educational institutions like campuses.
Enhancing GERD Disease Prediction using Extra Tree Classifier Tuned by Komodo Mlipir Algorithm Purba, Diya Namira; Fariani, Rida Indah
International Journal of Advances in Data and Information Systems Vol. 6 No. 3 (2025): December 2025 - International Journal of Advances in Data and Information Syste
Publisher : Indonesian Scientific Journal

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.59395/ijadis.v6i3.1428

Abstract

Gastroesophageal reflux disease (GERD) is a prevalent gastrointestinal disorder characterized by the backward flow of gastric contents into the esophagus, often causing heartburn and regurgitation, with a global prevalence of approximately 13.98%. Early detection is essential to prevent severe complications such as esophagitis, esophageal strictures, and esophageal cancer. However, conventional diagnostic methods are often limited by inadequate healthcare resources and high cost, particularly in developing countries. On the other hand, machine learning can be implemented as a promising alternative method for disease detection, improving accuracy through data pattern identification. Machine learning has been used for several disease detection tasks, such as Breast Cancer, Diabetes, etc. This study proposed an enhanced GERD prediction model by implementing the Extra Tree classifier optimized by the Komodo Mlipir Algorithm (KMA) for hyperparameter optimization.  This study used a GERD dataset from the Harvard  Dataverse, which consists of 1200 rows with 69 features. The result shows that the Extra Tree Algorithm that KMA tuned achieved a high-performance evaluation with an F1-score of 0.97.  This highlights the effectiveness of KMA in enhancing model performance. Compared to the previous study, the proposed Extra Tree Models optimized by KMA performed improved performance, demonstrating the effectiveness of metaheuristic optimization in GERD prediction.
Firefly Algorithm Under-sampling for Imbalance Data in Breast Cancer Survival Prediction Purba, Diya Namira; Namora Purba
Jurnal RESTI (Rekayasa Sistem dan Teknologi Informasi) Vol 10 No 1 (2026): February 2026
Publisher : Ikatan Ahli Informatika Indonesia (IAII)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.29207/resti.v10i1.6439

Abstract

Breast cancer remains a major health challenge, affecting approximately 1.7 million individuals annually and often leading to severe complications. Predicting survival outcomes is difficult due to highly imbalanced data, with 3,408 death cases compared to only 616 survival cases. To address this issue, we applied the Firefly Algorithm–based under-sampling (FAUS) to balance the dataset and combined it with three machine learning classifiers: Random Forest (RF), Decision Tree (DT), and K-Nearest Neighbor (KNN). Experimental results show that FAUS substantially improves predictive performance compared to conventional under-sampling. Among the tested models, RF achieved the highest F1-score of 0.79, while DT and KNN reached 0.72 and 0.68, respectively. The results indicate that FAUS is effective in preserving representative samples, thereby enhancing model performance in breast cancer survival prediction.