Skip to content

Machine learning-based fraud detection system capable of identifying and preventing fraudulent transactions in real-time for Finex, a financial service provider based in Florida.

Notifications You must be signed in to change notification settings

SowmyaKothari/Credit-Card-Fraud-Detection

Repository files navigation

Credit Card Fraud Detection: A Data Science Approach

Project Introduction

The rise in digital payment channels has significantly increased the risk of fraudulent transactions, posing a substantial threat to both financial institutions and their customers. Finex, a leading financial service provider based in Florida, has been facing a critical challenge due to an alarming rise in unauthorized credit and debit card transactions. These fraudulent activities have not only led to considerable financial losses but have also undermined customer trust and the bank’s credibility.

To address this issue, this project focuses on developing a machine learning-based fraud detection system capable of identifying and preventing fraudulent transactions in real-time. The project involves understanding the pipeline of a typical transaction, identifying the challenges at each step, and creating a robust solution that minimizes revenue loss while maintaining a seamless customer experience.

Project Summary

Business Problem

Finex has observed a significant increase in unauthorized credit and debit card transactions, resulting in severe financial losses. The bank's existing systems are not equipped to detect these fraudulent activities in a timely manner, leading to delayed responses and substantial payouts to affected customers. The goal of this project is to develop a machine learning model that accurately detects fraudulent transactions and reduces the financial impact on the bank.

Data Overview

The dataset used for this project includes approximately 1.85 million transactions from 1,000 cardholders across 800 merchants, collected between January 1, 2019, and December 31, 2020. Out of these transactions, 9,651 are labeled as fraudulent, making up 0.52% of the dataset. Given the highly imbalanced nature of the data, special attention is required to ensure the model’s accuracy in detecting fraud.

Project Pipeline

  1. Data Understanding: Load and analyze the dataset to identify key features and their distributions.
  2. Exploratory Data Analysis (EDA): Perform univariate and bivariate analyses, handle data skewness, and prepare the data for model building.
  3. Data Splitting: Split the data into training and testing sets using stratified k-fold cross-validation to ensure that the minority class is appropriately represented.
  4. Model Building & Hyperparameter Tuning: Develop various machine learning models, starting with baseline models and moving towards more complex ensembles, fine-tuning them for optimal performance.
  5. Model Evaluation: Evaluate the models using appropriate metrics, with a focus on accurately identifying fraudulent transactions. Metrics like precision, recall, and F1-score are considered, given the imbalanced nature of the data.
  6. Cost-Benefit Analysis: Perform a cost-benefit analysis to compare the costs incurred by the bank before and after deploying the fraud detection model. This involves calculating the total savings by reducing fraudulent transactions through the model's predictions.

Results

The developed fraud detection system provides a substantial reduction in the financial losses incurred due to fraudulent transactions. By implementing a secondary layer of authentication for transactions flagged as potentially fraudulent, the bank can significantly lower the number of undetected fraudulent transactions. The cost-benefit analysis demonstrates the long-term financial benefits of deploying this model, ensuring that Finex can safeguard its revenue and maintain customer trust.

Conclusion

This project showcases the practical application of machine learning in solving real-world financial problems. The proposed solution not only helps in detecting and preventing credit card fraud but also contributes to the broader goal of securing digital payment systems in an increasingly interconnected world.

About

Machine learning-based fraud detection system capable of identifying and preventing fraudulent transactions in real-time for Finex, a financial service provider based in Florida.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published