Credit Card Fraud Detection

Table of Contents

1. Project Overview
2. Dataset
3. Data Preprocessing
4. Model training
5. Model Evaluation
6. Results and Insights
7. How to run the project
8. Conclusion

1. Project Overview

The Project aims to detect whether a Credit card transaction is Legit or Fraudulent using Machine Learning Techniques in Python.

2. Dataset

Download the Dataset here : Credit card This Dataset contains:

• A large number of transactions with various features.

• As the credit card informations are highly sensitive the features are named as V1,V2...

• A highly unbalanced distribution of Legit vs. fraudulent transactions.

3. Data Preprocessing

• Loading the dataset using Pandas.

• Checking for missing values and handling them accordingly.

• As the dataset was highly unbalanced. To balance the ratio, the dataset was split into legit_transactions and fraud_transactions.

• Perform scaling and selection.

• Splitting the dataset into training and testing sets.

4. Model Training

• As the data can be classified into legit and fraud we will be using Logistic Regression as a baseline model.

• Other possible models that can be used: Decision Trees, Random Forest, and Neural Networks.

5. Model Evaluation

• The performance of the model is assessed using accuracy, precision, recall, and F1-score.

6. Results

• The model works perfectly as the training dataset accuracy score is not significantly larger or miniscule than the testing data set.

• The model does not have issue with underfitting or overfitting.

7. How to run the Project

1. Install the required dependencies using pip install -r requirements.txt.

2. Run the Jupyter Notebook step by step.

3. Evaluate the model’s performance and make necessary modifications.

8. Conclusion

This project demonstrates the effectiveness of machine learning techniques in detecting fraudulent transactions.