Abstract

In this thesis, real data from real fraud cases was leveraged to train a machine learning model. With the access to real data, model validation was conducted to prove the model was able to identify fraudulent transactions that it had never seen before leveraging patterns derived from Benford's Law. This law has proven powerful in recognizing patterns of numerical data that are un-natural. This set of work expands on the known powerful capabilities of Benford's Law by layering in related features that help to make the system able to recognize patterns in transactions very effectively. The overall objective is to focus large sets of data that contain very few fraudulent transactions down to a manageable number of high risk transactions to be further investigated.

Advisor

Visa, Sofia

Department

Computer Science

Disciplines

Computer Sciences | Data Science | Mathematics

Publication Date

2024

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Share

COinS
 

© Copyright 2024 Grant Mitchell