Published Online:April 2026
Product Name:The IUP Journal of Accounting Research & Audit Practices
Product Type:Article
Product Code:IJARAP060226
DOI:10.71329/IUPJARAP/2026.25.2.152-168
Author Name:Jeelan Basha V
Availability:YES
Subject/Domain:Finance
Download Format:PDF
Pages:152-168
The paper investigates the anomalies in Karnataka’s monthly Goods and Services Tax (GST) collections using a hybrid forensic framework combining Benford’s Law and the Isolation Forest algorithm. Monthly data for the period 2017-2024 were analyzed using computing software R. The findings revealed nonconformity in all GST components, particularly State GST (SGST) and CESS, with high MAD and distortion values. Isolation Forest flagged extreme anomalies using both fixed thresholds and contamination rates. Practical implications include targeted audits and integration of anomaly detection in tax monitoring. The study offers an original contribution by applying dual-layered statistical and machine learning (ML) techniques to public tax data for enhanced fraud detection and governance.
In an era of digital governance and data-driven compliance, forensic auditing has emerged as a critical tool in detecting and preventing financial irregularities in public revenue systems. In India, the introduction of the Goods and Services Tax (GST) marked a historic reform aimed at improving tax transparency, reducing evasion, and unifying the indirect tax structure. However, the complexity of GST reporting—spread across Central GST (CGST), State GST (SGST), Integrated GST (IGST) and CESS—has also opened new avenues for misreporting, data manipulation, and revenue leakage. As tax systems increasingly depend on automated processes and high-volume data submissions, the role of forensic audit tools such as Benford’s Law and machine learning (ML) models becomes ever more pertinent.