STATISTICAL APPLICATIONS IN FRAUD DETECTION: UNCOVERING HIDDEN PATTERNS IN FINANCIAL STATEMENTS DURING AUDITS

This study explores the application of statistical techniques in fraud detection, aiming to uncover hidden patterns in financial statements during audits. Using a mixed-methods approach, the research analyzed audit reports from 2020 to 2024, applying regression analysis, Chi-Square tests, and anomaly detection techniques. The findings reveal a significant increase in fraud detection accuracy, with machine learning models improving from 75% in 2020 to 95% in 2024. Regression analysis showed a strong positive correlation (R² = 1.0, p < 0.001) between financial ratio deviations and fraud probability, with a 30% increase in fraud likelihood per unit deviation. Chi-Square results (χ² = 5.62, p = 0.131) confirmed that fraud detection aligns with expected fraud distributions. Additionally, fraud detection advancements correlated inversely with audit duration (-1.0), reducing the average audit time from six months in 2020 to five months in 2024. The study concludes that integrating statistical tools enhances fraud detection efficiency, but challenges such as data quality and technological resistance persist. Recommendations include auditor training in statistical methods, hybrid fraud detection models, improved data validation, real-time monitoring, and cross-disciplinary collaboration to optimize fraud detection practices.