Kent0625/Intelligent_Mobile_Wallet_Fraud_Detection_System

Intelligent Mobile Wallet Fraud Detection System

A real-time fraud detection pipeline combining Machine Learning with Generative AI forensics for automated transaction analysis.

[image]

Overview

This system processes mobile money transactions to detect fraudulent activity in real-time. It uses a supervised learning model to flag suspicious transactions and a Large Language Model (LLM) to provide an explainable forensic report for analysts.

Key Features

• —Real-time Inference: Low-latency detection using a specialized Random Forest model.
• —Generative AI Forensics: Automated narrative analysis of fraud using Hugging Face's Inference API (Qwen 2.5).
• —Explainable AI (XAI): SHAP (SHapley Additive exPlanations) values visualize specific feature contributions for every decision.
• —Production Pipeline: Robust stateful preprocessing using OneHotEncoder and StandardScaler to handle live data streams.

Technical Architecture

• —Model: Random Forest Classifier (Scikit-learn) trained on the PaySim dataset.
• —LLM Integration: huggingface_hub InferenceClient (Qwen/Qwen2.5-7B-Instruct).
• —Frontend: Streamlit for live monitoring and alert visualization.
• —Evaluation: Optimized for AUPRC (Area Under Precision-Recall Curve) on highly imbalanced data (1:1000).

Quick Start

1. Installation

git clone https://github.com/Kent0625/Intelligent-Mobile-Wallet-Fraud-Detection-System-Real-time-Detection-Generative-AI-Forensics.git
cd Intelligent-Mobile-Wallet-Fraud-Detection-System-Real-time-Detection-Generative-AI-Forensics
pip install -r requirements.txt

2. Configuration

Create a .env file with your Hugging Face token (required for AI analysis):

HF_TOKEN=your_token_here

3. Run Application

streamlit run src/dashboard.py

Project Structure

• —src/: Source code for data pipelines, feature engineering, and the dashboard.
• —models/: Serialized model artifacts (.pkl).
• —tests/: Unit tests for schema validation and pipeline integrity.