Andhika Seno Tamtama, 4611418019 (2022) PENINGKATAN AKURASI ALGORITMA RANDOM FOREST MENGGUNAKAN PCA DAN TEKNIK RESAMPLING DENGAN DATA AUGMENTATION UNTUK DETEKSI PENIPUAN TRANSAKSI KARTU KREDIT. Under Graduates thesis, Universitas Negeri Semarang.

PDF - Published Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Credit card fraud often results in significant financial damage. Specifically, the handling of credit card fraud in Indonesia is still less effective because legislation that does not set out specifically for those who have violated the rule and have not optimized the implementation of the handling done by law enforcement. Thus, a detection method must continue because no one can predict when fraud will occur even if there is already protection against it. Therefore, the study aims to detect credit card transactions whether transactions performed by users include fraudulent transactions or normal transactions. The detection method is to distinguish between a normal transaction and a fraudulent one. The credit�card transaction analysis uses a random forest algorithm as an algorithm for the classification process. The issue faced from the classification process by using credit card fraud fraud filing data sourced from Kaggle's source data fraud is an imbalanced data that causes an imbalanced value of data to be balanced on the model results from data training. To resolve the problem, it was used a combination of PCA methods and resampling techniques with augmentation data for the optimum process on random forest classification algorithms. The PCA method is used in the preprocessing stage as to do the process of transforming data into numerical data and resampling techniques and augmentation data are used for data resampling processes to bring the data to a balance. The data used is a data card fraud of Europe that has 284807 transactions. Model accuracy measurement was implemented using a confusion matrix. The highest accuracy results from a random forest combination using PCA and resampling techniques with a augmentation data of 99.9976%.

Actions (login required)

View Item