Vendor Invoice Matching & Anomaly Detection

Overview

This project aims to match vendor invoices with purchase orders (POs) and detect anomalies using NLP and anomaly detection techniques.

• Automated Invoice Matching using text similarity methods (TF-IDF, Levenshtein, BERT).
• Fraud & Anomaly Detection using machine learning models (Isolation Forest, Autoencoders).
• Data Visualization using Streamlit and Matplotlib.

Project Structure

• data/ - Contains the dataset (synthetic_vendor_invoices.csv).
• notebooks/ - Jupyter Notebooks for data preprocessing & model development.
• src/ - Python scripts for dataset generation, matching, and anomaly detection.