Article Per Year (5 Year)
p-Index From 2020 - 2025
0.23
P-Index
This Author published in this journals
All Journal METIK JURNAL
Rahman, Irfan Fauzia
Unknown Affiliation
Computer Science & IT Control & Systems Engineering Decision Sciences, Operations Research & Management Earth & Planetary Sciences Electrical & Electronics Engineering

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

 1 
Deteksi Struktur Material Perovskit ABO3 Berbasis Machine Learning Rahman, Irfan Fauzia; Al Azies, Harun; Akrom, Muhamad
METIK JURNAL (AKREDITASI SINTA 3) Vol. 9 No. 1 (2025): METIK Jurnal
Publisher : LPPM Universitas Mulia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47002/metik.v9i1.1036

Abstract

This study proposes a machine learning-based classification approach to identify perovskite and non-perovskite structures in ABO? compounds. Perovskites have garnered significant attention as a source of functional materials, including solar cells and catalysts. Yet, discovering new materials remains a considerable challenge in terms of efficiency and exploration speed. This research addresses this gap by offering a data-driven method that automatically classifies compound structures based on crystallographic and chemical descriptors. The dataset comprises various structural and chemical features, which are analyzed using descriptive statistics, boxplot visualization, and multivariate correlation to understand the data distribution and inter-feature relationships. Four machine learning algorithms, LightGBM, XGBoost, CatBoost, and K-Nearest Neighbors (KNN), were tested and evaluated based on accuracy, precision, recall, and F1 score. Results show that LightGBM achieved the best performance with 97% accuracy, a 98% F1 score, and a confusion matrix indicating minimal classification errors. Feature importance analysis identified the tolerance factor (t), the B to O atomic radii ratio, and the AO and BO bond lengths as the most influential features. These findings highlight that tree-based boosting models effectively capture complex structural patterns, and this approach can accelerate the discovery of new materials.
Co-Authors Akrom, Muhamad Harun Al Azies