Implementasi Extreme Learning Machine dan Explainable Artificial Intelligence Pada Deteksi Fraud Transaksi Pembayaran Online PT. XYZ

Pudyastuti, Radya Ardi Ninang (2025) Implementasi Extreme Learning Machine dan Explainable Artificial Intelligence Pada Deteksi Fraud Transaksi Pembayaran Online PT. XYZ. Undergraduate thesis, UPN Veteran Jawa Timur.

This is the latest version of this item.

	Text (Cover) 21083010097-Cover.pdf Download (1MB)
	Text (Bab 1) 21083010097-Bab 1.pdf Download (240kB)
	Text (Bab 2) 21083010097-Bab 2.pdf Restricted to Repository staff only until 17 September 2028. Download (559kB)
	Text (Bab 3) 21083010097-Bab 3.pdf Restricted to Repository staff only until 17 September 2028. Download (418kB)
	Text (Bab 4) 21083010097-Bab 4.pdf Restricted to Repository staff only until 17 September 2028. Download (1MB)
	Text (Bab 5) 21083010097-Bab 5.pdf Download (250kB)
	Text (Daftar Pustaka) 21083010097-Daftar Pustaka.pdf Download (235kB)
	Text (Lampiran) 21083010097-Lampiran.pdf Restricted to Repository staff only Download (383kB)

Abstract

The digital transformation in the era of Society 5.0 has brought significant changes to the financial sector through the adoption of financial technology. While offering convenience, this development has also triggered an increased risk of digital fraud. PT. XYZ, as a payment service provider, still relies on rule-based methods that are limited in addressing new and more complex fraud patterns. This study aims to develop a more accurate and transparent fraud detection system by implementing the Extreme Learning Machine (ELM) algorithm combined with Explainable Artificial Intelligence (XAI) using LIME. ELM was chosen due to its simple architecture with only one hidden layer, allowing for fast and efficient training on large-scale data. The model was tested under two scenarios, namely without resampling and with ADASYN, using fixed parameters: 100 neurons in the hidden layer, sigmoid activation function, and seed weight initialization of 42. The evaluation results show that the non-resampling scenario achieved an accuracy of 78.1%, precision of 98.3%, recall of 78.8%, and F1-score of 87.4%. After applying resampling with ADASYN, the performance improved, reaching an accuracy of 82.9%, precision of 97.8%, recall of 84.2%, and F1-score of 90.5%. These findings demonstrate that ELM is effective in detecting fraud, while the use of the ADASYN resampling technique enhances the model’s sensitivity to minority classes. Furthermore, the integration of LIME provides transparency by explaining the contribution of each feature to the predictions. Overall, this study highlights the effectiveness of ELM as the core of a fraud detection system, with resampling and XAI support that strengthen both accuracy and interpretability of the model.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Idhom, Mohammad	NIDN0010038305	idhom@upnjatim.ac.id
Thesis advisor	Hindrayani, Kartika Maulida	NIDN0009099205	kartika.maulida.ds@upnjatim.ac.id

Subjects:

Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Q Science > QA Mathematics > QA76.6 Computer Programming

Divisions:

Faculty of Computer Science > Departemen of Data Science

Depositing User:

Radya Ardi Ninang

Date Deposited:

19 Sep 2025 03:18

Last Modified:

19 Sep 2025 03:18

URI:

https://repository.upnjatim.ac.id/id/eprint/43831