Article Per Year (5 Year)
p-Index From 2021 - 2026
0.408
P-Index
This Author published in this journals
All Journal Indonesian Journal of Artificial Intelligence and Data Mining Journal of Computing Theories and Applications
Muhammad Fahmy Nadhif
Sebelas Maret University
Computer Science & IT Decision Sciences, Operations Research & Management

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Evaluating Single and Hybrid Feature Selection for Rainfall Prediction Using XGBoost Bambang Widoyono; Muhammad Fahmy Nadhif; Ridha Adjie Eryadi
Indonesian Journal of Artificial Intelligence and Data Mining Vol. 9 No. 1 (2026): March 2026
Publisher : Universitas Islam Negeri Sultan Syarif Kasim Riau

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

Abstract

Rainfall prediction is challenging due to the complex and nonlinear nature of meteorological data. Previous studies using XGBoost with feature selection have demonstrated superior performance compared to other models, but evaluations have focused solely on error metrics (RSME, SME, MAE). Recent research suggests that predictive models should be evaluated for generalization, stability, interpretability, and computational efficiency to ensure their reliability. To close this gap, this study uses 8,750 hourly records obtained from Open-Meteo with 81 engineered features to evaluate XGBoost under three scenarios: without feature selection, single feature selection (MI, Boruta, SHAP, mRMR, ReliefF), and hybrid feature selection. Our findings demonstrate that accuracy is not always increased by feature selection. It does, however, increase interpretability, decrease overfitting, and improve computational efficiency. SHAP provides the most reliable performance among single methods, achieving lower RMSE (0.72632) and improved stability. Hybrid feature selection produces the most balanced performance gap = 0.01325, and stable variance = 0.03315 while reducing feature complexity to 35 variables. This study theoretically shows the value of multidimensional evaluation that goes beyond error metrics. In practical terms, this study suggests a feature selection method for rainfall prediction systems that are effective, reliable, and simple to understand.
Evaluating Disruption Recovery across Pareto-Representative Solutions in Multi-Objective University Course Timetabling Hanifah Salsabila Ryadi; Ristu Saptono; Muhammad Fahmy Nadhif
Journal of Computing Theories and Applications Vol. 4 No. 1 (2026): JCTA 4(1) 2026
Publisher : Universitas Dian Nuswantoro

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.62411/jcta.17110

Abstract

Multi-objective university course timetabling produces a Pareto front of trade-off solutions, but ultimately, one timetable must be selected for deployment. Existing studies commonly base this choice on objective values, robustness criteria, or stakeholder preferences, while the downstream recovery implications of selecting different Pareto representatives remain insufficiently understood. This study proposes a deployment-oriented evaluation framework to examine whether the selected Pareto-representative timetable affects recovery after calendar-based disruptions. The framework keeps the original optimization objectives fixed, selects the best-f1, best-f2, and compromise timetables from each run, projects them onto an academic calendar with actual dates and holidays, and evaluates them under five disruption scenarios using greedy direct rescheduling followed by target-meeting fulfillment repair. The framework is evaluated using institutional scheduling and calendar data from the odd and even semesters of the 2025/2026 academic year, which provide a natural contrast in room-slot occupancy and disruption exposure. Weekly timetables are generated using NSGA-II by minimizing a weighted soft-constraint penalty (f1) and average room-capacity waste (f2), with the experiments repeated across five independent random seeds. No representative timetable produced a consistent recovery advantage under the tested setting: differences in the direct rescheduling success rate were small relative to seed-to-seed variation, even when the representative timetables differed substantially in their class placements. The larger contrast occurred between semesters. The denser odd semester, with approximately 91% room-slot occupancy, directly recovered about one-third of affected meetings, whereas the less dense even semester, with approximately 52% occupancy, directly recovered nearly all affected meetings. Both semesters achieved 100% target-meeting fulfillment after repair. Overall, the stronger observed contrast occurred between semester settings and was more consistent with differences in structural slack than with the selected representative category.
Co-Authors Bambang Widoyono Hanifah Salsabila Ryadi Ridha Adjie Eryadi Ristu Saptono