Pipeline Machine Learning Operations Berbasis Low-Code Untuk Deployment Dan Monitoring Model Deteksi Penipuan Kartu Kre

Syahputra, Ahmad Wildan (2026) Pipeline Machine Learning Operations Berbasis Low-Code Untuk Deployment Dan Monitoring Model Deteksi Penipuan Kartu Kre. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5024221064-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Pengungkapan lingkungan oleh emiten menjadi aspek penting dalam mendukung transparansi keberlanjutan, khususnya pascaimplementasi regulasi keuangan berkelanjutan di Indonesia. Namun, peningkatan jumlah laporan belum sepenuhnya mencerminkan kualitas dan keseimbangan antar dimensi lingkungan. Penelitian ini bertujuan untuk mengukur tingkat pengungkapan lingkungan emiten Indonesia periode 2018–2023 melalui Environmental Disclosure Score (ENV) dan Disclosure Consistency Index (DCI), menganalisis tren perkembangan dan konsistensi pengungkapan antar subpilar lingkungan, serta mengevaluasi perbedaan sektoral dan tipologi emiten berdasarkan karakteristik pengungkapan lingkungan. Metode penelitian menggunakan pendekatan kuantitatif deskriptif-eksploratif berbasis data panel dari ESG Intelligence (ESGI). Analisis dilakukan melalui perhitungan Environmental Disclosure Score (ENV), Disclosure Consistency Index (DCI) berbasis modifikasi konsep Coefficient of Variation (CV), analisis tren menggunakan Compound Annual Growth Rate (CAGR), analisis korelasi antar subpilar, uji Kruskal–Wallis dan Dunn Test, serta pengelompokan emiten menggunakan metode K-Means Clustering. Hasil penelitian menunjukkan bahwa rata-rata Environmental Disclosure Score (ENV) meningkat dari 0,216 pada tahun 2018 menjadi 0,734 pada tahun 2023, terutama pada subpilar energi, limbah, dan emisi karbon. Sebaliknya, nilai rata-rata Disclosure Consistency Index (DCI) menurun dari 0,835 menjadi 0,747, yang menunjukkan bahwa peningkatan tingkat pengungkapan lingkungan belum sepenuhnya diikuti oleh konsistensi pengungkapan antar subpilar lingkungan. Analisis sektoral menunjukkan adanya perbedaan signifikan tingkat pengungkapan lingkungan antar sektor industri. Selanjutnya, hasil K-Means Clustering mengelompokkan 33 emiten ke dalam tiga tipologi, yaitu Cluster 1 (Transparan & Konsisten) sebanyak 13 emiten, Cluster 2 (Selective Discloser) sebanyak 11 emiten, dan Cluster 3 (Insufficient Discloser) sebanyak 9 emiten. Berdasarkan karakteristik masing-masing klaster serta subpilar lingkungan yang masih memiliki tingkat pengungkapan relatif rendah, disusun rekomendasi untuk meningkatkan pengelolaan dan pengungkapan aspek air serta biodiversitas. Secara keseluruhan, transparansi lingkungan emiten di Indonesia menunjukkan peningkatan, namun karakteristik pengungkapan lingkungan masih bervariasi antar emiten.
=================================================================================================================================
The growth of digital banking transactions has increased the risk of credit card fraud. Although machine learning detection models may perform well during training, their performance can degrade in production due to concept drift, while manual deployment processes remain slow and prone to human error. This research designs and implements a low-code Machine Learning Operations (MLOps) pipeline using the n8n platform for the deployment and monitoring of an XGBoost-based credit card fraud detection model. The proposed system integrates a GitHub-based model registry with a hot-reload mechanism that enables zero-downtime model updates, prediction logging for improved observability, and statistical drift detection using the Kolmogorov–Smirnov test for data drift and the Chi-Squared test with Fisher’s Exact Test as a fallback for prediction drift. The novelty of this research lies in a two-tier alerting scheme that applies stricter thresholds to the three most important model features, thereby reducing false alarms caused by multiple statistical testing. Experimental results show that the system successfully served all 960 requests during model updates without failure, achieved a hot-reload time of 179.9 milliseconds, accurately detected synthetic drift without generating false alarms on normal data, and reduced deployment lead time by 96.6%, from 21.36 seconds to 0.72 seconds, while maintaining near-zero median logging overhead. These findings demonstrate that a low-code MLOps approach is feasible for improving the reliability and operational efficiency of credit card fraud detection systems.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Penipuan Kartu Kredit, Machine Learning Operations (MLOps), Low-Code, Concept Drift, Deployment Otomatis, Monitoring Model, Credit Card Fraud, Machine Learning Operations (MLOps), Low-Code, Concept Drift, Automated Deployment, Model Monitoring.
Subjects:	T Technology > T Technology (General) > T58.5 Information technology. IT--Auditing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Ahmad Wildan Syahputra
Date Deposited:	22 Jul 2026 06:41
Last Modified:	22 Jul 2026 06:41
URI:	http://repository.its.ac.id/id/eprint/136236

Available Versions of this Item

• Pipeline Machine Learning Operations Berbasis Low-Code Untuk Deployment Dan Monitoring Model Deteksi Penipuan Kartu Kre. (deposited 22 Jul 2026 06:41) [Currently Displayed]
  • Pipeline Machine Learning Operations Berbasis Low-Code Untuk Deployment Dan Monitoring Model Deteksi Penipuan Kartu Kredit. (deposited 23 Jul 2026 03:29)

Actions (login required)

View Item