🚀 Vendor Invoice Intelligence System

📌 Table of Contents

• Project Overview
• Business Objective
• Data Source
• Exploratory Data Analysis (EDA)
• Models Used
• Evaluation Metrics
• Application
• Project Structure
• How to Run This Project
• Author

---

📖 Project Overview

This project implements an end-to-end Machine Learning system designed to support finance teams by:

• Predicting expected freight costs for vendor invoices
• Flagging high-risk invoices that require manual review

The system helps detect abnormal cost, freight, and operational patterns.

---

🎯 Business Objective

• Invoice Cost Leakage: Minimize unnecessary expenses
• Audit Risk Management: Ensure compliance and financial accuracy

---

🗄️ Data Source

Data is stored in a relational SQLite database containing:

• Vendor Invoice Table
• Purchase Table
• Purchase Prices Table
• Inventory Table

---

📊 Exploratory Data Analysis (EDA)

• Analyzed cost distribution and anomaly patterns
• Performed feature engineering
• Used statistical tests (T-test) to validate feature significance

---

🤖 Models Used

Regression (Freight Cost)

• Linear Regression
• Decision Tree
• Random Forest

Classification (Invoice Risk)

• Logistic Regression
• Random Forest (final model)
  • Tuned using GridSearchCV
  • Handles class imbalance

---

📏 Evaluation Metrics

Regression

• RMSE
• R² Score

Classification

• Precision
• Recall
• F1 Score

---

💻 Application

• Built using Streamlit
• Provides a user-friendly interface for:
  • Inputting invoice data
  • Getting real-time predictions
  • Flagging risky invoices

---

📁 Project Structure

├── 📁 freight_cost_prediction
│   ├── 📁 models
│   ├── data_preprocessing.py
│   ├── model_evaluation.py
│   └── train.py
├── 📁 invoice_flagging
│   ├── data_preprocessing.py
│   ├── model_evaluation.py
│   └── train.py
├── 📁 inferencing
│   ├── predict_freight.py
│   └── predict_invoice_flag.py
├── 📁 models
│   ├── predict_flag_invoice.pkl
│   ├── predict_freight_model.pkl
│   └── scaler.pkl
├── 📁 notebooks
│   ├── flagVendorInvoice.ipynb
│   └── freightCost.ipynb
├── .gitignore
├── app.py
└── requirements.txt

---

🛠️ How to Run This Project

Clone the Repository

git clone https://github.com/Prashantbhati7/invoice-intelligence.git
cd invoice-intelligence

--

Train Models

Freight Cost Model

python freight_cost_prediction/train.py

Invoice Risk Model

python invoice_flagging/train.py

---

5. Run the Application

streamlit run app.py

---

⚠️ Notes

• Make sure .pkl model files are generated before running the app
• If errors occur, verify dependencies installation
• Use a virtual environment to avoid conflicts

---

👤 Author

Prashant Bhati