An open API service indexing awesome lists of open source software.

https://github.com/eva-kaushik/fraud-detection-mlops

An AI-powered application leveraging machine learning for detecting fraudulent activities in financial transactions with MLOps.
https://github.com/eva-kaushik/fraud-detection-mlops

financial-analysis fraud-detection fraud-prevention h2o-automl h2oai mlops tensorflow

Last synced: 3 months ago
JSON representation

An AI-powered application leveraging machine learning for detecting fraudulent activities in financial transactions with MLOps.

Awesome Lists containing this project

README

          

# Fraud-Detection-MLOps
## Fraud Detection System

## Overview
The Fraud Detection System is an AI-powered application designed to detect fraudulent activities in financial transactions. It leverages machine learning techniques to analyze historical data and predict the likelihood of a transaction being fraudulent.

## Features
- **Data Collection**: Collects data from NSFAS and University of Limpopo CSV files.
- **Data Preprocessing**: Cleans, preprocesses, and merges the data, handling missing values and performing feature engineering.
- **Model Training**: Trains a neural network classifier using TensorFlow/Keras.
- **Model Evaluation**: Evaluates the trained model's performance using accuracy, precision, recall, and F1 score.
- **SMS Confirmation**: Generates a random confirmation code and sends it via SMS to the user for confirmation.

## Requirements
- Python 3.x
- Pandas
- Scikit-learn
- Imbalanced-learn
- Keras
- TensorFlow
- H2O.ai
- Other dependencies listed in `requirements.txt`

## Usage
Installation: Install the required packages listed in `requirements.txt`.
```bash
pip install -r requirements.txt
Data Preparation: Ensure that the NSFAS and University of Limpopo CSV files are available and correctly located. Update the file paths in the code if necessary.
Running the Application: Execute the main script Fraud Detection.py.
bash
python fraud_detection.py
Result Analysis: Check the output for model evaluation metrics and the confirmation code sent via SMS.