Date of Award
2021
Publisher
North Dakota State University
Document Type
Master's Paper
Degree Awarded
Master of Science (MS)
Program
Computer Science
Department
Computer Science
College
Engineering
Faculty Advisor
Ludwig, Simone
Description
Technology is advancing rapidly, and more tasks are becoming online than ever. Along with the benefits comes the disadvantages of this great advancement. While online services relieve from the struggle of in person activities, it also puts you on the risk of getting deceived by the fraudsters. This paper aims to detect the fraudulent transactions made online from a bank using a synthetically produced dataset. A random forest model has been trained to predict the fraudulent transactions. To achieve the best performance, the hyperparameters of the model have been tuned using three different tuning methods. As it turns out, grid search proved to be the best tuning strategy in terms of the mean cv score, precision, recall, f1-score and accuracy. It only lacked in providing the best run time, where Bayesian Optimization scored well than the others.
Recommended Citation
Sarao, Kulwinder Kaur, "Analyzing Three Different Tuning Strategies for Random Forest Hyperparameters for Fraud Detection" (2021). Computer Science. 311.
https://digitalcommons.ndsu.edu/computer-sci/311
Rights
NDSU policy 190.6.2
Rights Link
https://www.ndsu.edu/fileadmin/policy/190.pdf
Handle Identifier
https://hdl.handle.net/10365/32253