This project aims to develop a machine learning model for predicting fraudulent transactions within a financial company. The goal is to not only identify potentially fraudulent activities but also to derive actionable insights from the model to enhance fraud prevention strategies.The dataset used for this project consists of 6,362,620 rows and 10 columns, sourced from Kaggle. It includes various types of transactions, such as CASH-IN, CASH-OUT, DEBIT, PAYMENT, and TRANSFER. Notably, fraudulent transactions are primarily found in the TRANSFER and CASH-OUT categories, with a significant imbalance between genuine and fraudulent transactions—approximately 8000 fraudulent transactions out of 2.8 million relevant entries.