Financial Anomalies and Creditworthiness: A Python-Driven Machine Learning Approach Using Mahalanobis Distance for ISE-Listed Companies in the Production and Manufacturing Sector

Abstract

This paper investigated the effect of financial ratio anomalies on the creditworthiness of companies in the production and manufacturing sectors listed on the Istanbul Stock Exchange. Financial ratios are crucial in assessing creditworthiness as they reflect potential financial risks, but anomalies in these ratios can skew credit evaluations. The statistical analysis is carried out using the Mahalanobis distance, a Python-driven machine learning technique, with financial data from 254 companies over the period of 2014 to 2023. The findings show that 89.61% of the observations were identified as normal, falling into low and moderate risk categories, whereas 10.39% were considered anomalies or high-risk, indicating that a substantial proportion of companies maintained financial indices within expected ranges. Besides the study observed significant risk category variations, particularly in the latter half of each year, likely reflecting differing reporting and auditing practices. Farther, Principal Component (PCA) analysis indicated that Asset Profitability, EBITDA to Total Debts, Short-Term Debt to Current Assets, Net Sales to Short-Term Debts, and Current Ratio are the most influential financial ratios. On the other hand, the least influential financial ratios in PCA include Net Working Capital Turnover, Operating Expenses to Average Assets, Market Value to EBITDA, Efficiency Ratio (Assets Turnover Ratio), and Change in Depreciation Expenses Ratio (annual difference in depreciation expense). In conclusion, this study underscores the complex role of financial ratio anomalies in Turkish industries in assessing creditworthiness.