💼 AuditIQ — AI-Powered Fraud Detection

🚀 Live Demo: https://auditiq1.streamlit.app/
📊 GitHub Repo: github.com/kaushikatla-cell/AuditIQ

---

🧠 Overview

AuditIQ scans accounting data to automatically detect irregular or high-risk transactions using unsupervised machine-learning models.
It helps users quickly identify potential fraud or anomalies across their financial data.

Ideal for:

• Students learning accounting or data analytics
• Auditors or finance interns reviewing transactions
• Small businesses tracking unusual activity

---

⚙️ Features

✅ Upload CSV data (Date, Description, Category, Amount)
✅ Detect anomalies using Isolation Forest ML model
✅ Interactive data visualizations (Seaborn + Matplotlib)
✅ Adjustable anomaly sensitivity
✅ Downloadable “Audit Report” with risk flags
✅ Streamlit-powered, no installation required

---

🧩 Tech Stack

• Python 3.10+
• Streamlit — frontend dashboard
• Pandas & NumPy — data handling
• Scikit-learn — anomaly detection
• Matplotlib / Seaborn — visualization

---

📂 Project Structure

---

🧪 How It Works

1. Upload or use the included sample_data.csv
2. AuditIQ processes the dataset and computes anomaly scores
3. Suspicious transactions are flagged visually and listed in a report
4. Adjust sensitivity using the sidebar to refine detection results

---

📸 Preview

Dashboard Preview

(You can upload your own screenshot to Imgur and replace the link above)

---

🧠 Example Use Case

Scenario:
A freelance accountant uploads a client’s expense data.
AuditIQ highlights one suspicious “Client Payment - Income - $12,000” transaction — potentially miscoded or fraudulent.

---

🧰 Local Setup

If you want to run it locally:

git clone https://github.com/kaushikatla-cell/AuditIQ.git
cd AuditIQ
pip install -r requirements.txt
streamlit run app.py