Anomaly Detection in Corporate Balance Sheets for Financial Risk Assessment Using Isolation Forest from 2020 to 2023

Abstrak

This study aims to evaluate corporate financial risk by analyzing changes in balance sheet accounts from 2020 to 2023 using anomaly detection methods. Employing the Isolation Forest algorithm with a 5% contamination rate, we identified a consistent 3,264 anomalies each year out of a total of 65,296 entries, focusing on key accounts, including Accumulated Depreciation (61 anomalies), Additional Paid-In Capital (17 anomalies), Accounts Payable (9 anomalies), and Accounts Receivable (6 anomalies). These anomalies highlight areas of potential financial risk associated with asset valuation, capital structure, and cash flow management. The steady presence of anomalies suggests underlying, possibly systemic factors influencing financial stability. Findings indicate that significant fluctuations in Accumulated Depreciation and Additional Paid-In Capital may impact the company’s asset valuation and investor perceptions, while irregularities in Accounts Payable and Accounts Receivable suggest short-term liquidity risks. Recommendations include regular monitoring of high-risk accounts, trend analysis to identify cyclical patterns, and examining correlations with macroeconomic conditions to understand root causes. Future research should explore advanced anomaly detection models and integrate real-time detection capabilities to enhance proactive financial risk management. This study demonstrates the effectiveness of anomaly detection in identifying critical financial risks, supporting improved decision-making and corporate resilience