Fraud Warden

Overview

Fraud Warden is a next-generation credit card fraud detection system that uses machine learning to predict whether a transaction is fraudulent or not. The system leverages a Random Forest Classifier to make predictions based on various features of the transaction.
Technology Stack Programming Language: Python

Libraries:

• streamlit for building the web application
• pandas for data manipulation
• plotly.express and seaborn for data visualization
• scikit-learn for machine learning
• pickle for model serialization

Installation Instructions

• Clone the Repository:
• git clone https://github.com/yourusername/fraud-warden.git
• cd fraud-warden
• Create a Virtual Environment:
  • python -m venv venv
  • source venv/bin/activate # On Windows use venv\Scripts\activate
• Install Dependencies:
  • pip install -r requirements.txt
• Run the Application:
  • streamlit run app.py

How It Works

• Data Preprocessing:
  • The application preprocesses the uploaded CSV file by removing unnecessary columns and converting date columns to datetime objects. Additional features such as time_of_day and age are derived from existing columns.
• Feature Engineering:
  • Categorical features are encoded into numerical values. The data is reindexed to ensure all required columns are present.
• Oversampling:
  • The application uses Synthetic Minority Over-sampling Technique (SMOTE) to balance the dataset.
• Model Prediction:
  • The preprocessed data is fed into a pre-trained Random Forest Classifier model. The model predicts whether a transaction is fraudulent based on the input features.
• Visualization:
  • The application provides various visualizations such as histograms, bar charts, and correlation heatmaps to help users understand the data.

Features

• Upload CSV: Users can upload a CSV file containing transaction data.
• Data Preview: Displays a preview of the uploaded data.
• Basic Statistics: Shows basic statistics of the dataset.
• Data Types: Displays the data types of each column.
• Missing Values: Shows the count of missing values in each column.
• Distribution of Numerical Columns: Visualizes the distribution of numerical columns.
• Counts of Categorical Columns: Visualizes the counts of categorical columns.
• Correlation Heatmap: Displays a heatmap of the correlation between numerical features.
• SMOTE Sampling: Balances the dataset using SMOTE sampling.
• Fraud Prediction: Predicts whether a transaction is fraudulent based on user input.

Resources Used

• Dataset: Credit Card Fraud Detection Dataset (Kaggle)
• Sklearn Documentation: Random Forest Classifier
• Streamlit Documentation: Streamlit
• Plotly Documentation: Plotly Express
• Seaborn Documentation: Seaborn
• Pandas Documentation: Pandas
• Python Documentation: Python
• SMOTE Documentation: SMOTE