Financial Forensics

A course on detecting financial fraud using both traditional forensic accounting techniques and modern data-driven methods like anomaly detection and data visualization.

This course provided a unique blend of traditional accounting and modern data science to uncover financial fraud. We began with the fundamentals of financial statements and common fraud vulnerabilities. The curriculum then introduced the formal role and process of forensic accounting, including practical, hands-on techniques like applying Benford’s Law, aging analysis, and outlier detection in Excel. The second half of the course shifted to a high-tech approach, exploring how AI/ML is used in the financial transaction lifecycle, and delving into supervised and unsupervised anomaly detection models to identify suspicious activities. The course culminated in learning how to visualize and explain these findings using Tableau.

Instructor

Dr. Arun Kumar G, Department of Management Studies (DOMS), IIT Madras

Course Schedule & Topics

The course is structured over 12 weeks, bridging the gap between classical accounting and modern data-driven forensic techniques.

Week Primary Focus Key Topics Covered
1 Introduction to Accounting & Finance Financial Statements, their relevance, and their role in business decision-making.
2 Fraud Vulnerabilities & Finance Types of fraud, their causes, common occurrences, and detecting red flags.
3 Forensic Accounting - 1 Introduction to forensic accounting, source of assignments, and the role of the expert.
4 Forensic Accounting - 2 The scope of forensic accounting work and the analytical process involved.
5 Benford’s Law & Implementation Understanding Benford’s Law and its practical implementation in Excel to detect fraud.
6 Analytical Detection Techniques Using Aging Analysis, Pareto Charts, and outlier identification to detect fraud.
7 Financial Transaction Lifecycle The financial transaction lifecycle and the application of AI/ML in modern finance.
8 Entity Resolution Understanding entities within transactions and the process of entity resolution.
9 Supervised Anomaly Detection Using labeled data to build models that detect anomalous transactions.
10 Unsupervised Anomaly Detection Using unlabeled data to identify unusual patterns and outliers in financial data.
11 Time-Based Anomaly Detection Techniques for detecting anomalies in time-series financial data.
12 Explainability & Visualization Model explainability and using Tableau for financial data visualization.

Material used