PERANCANGAN SISTEM VERIFIKASI OTOMATIS UNTUK LAPORAN PENIPUAN DI PT KREDIBEL TEKNOLOGI INDONESIA MENGGUNAKAN TEKNIK DATA MINING

• Main Author: Muhamad Iqbal, Fadel
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55730
• Institution: Institut Teknologi Bandung

PT Kredibel Teknologi Indonesia is a private company that is focused on online fraud issues. The company receives fraud report from the community and use that report to measure someone credibility score. In 2020, the company launched a new product, Fraud Management System, that has a function to help fintech companies identify their new customers. Nowadays, the company verifies every fraud report manually and the productivity of that verification process depends on its verificators. The time spent on this process is approximately more than 1 minute. To solve that problem, the company wants to create an automation system for the verification process using data mining. This research uses 11 relevant and available variables with 27,238 data sets which come from fraud report that has been manually verified in the past. The historical reports are used for the training model. This research uses Decision Tree dan Gradient Boosting Tree (Extreme Gradient Boosting dan Light Gradient Boosting Machine type) as a classification model. To make sure that the model can create a good performance measure, this research uses two series process. The first process is used for classifying fraud or non-fraud chronology. The second prose is used for the whole verification system. These combined models create verification time reductions from 61.4 second/report to 0.04 second/report with 93% accuracy for validation data. Therefore, the application is made to help the company verify fraud reports automatically.