المؤسسات الصغيرة والمتوسطة بين إشكالية التمويل واحتمال التعثر

Abstract

Early prediction of financial distress is a cornerstone for ensuring the continuity of small and medium-sized enterprises (SMEs) in volatile economic environments. This study evaluates the effectiveness of artificial intelligence (AI) models, specifically Deep Neural Networks (DNN)and Genetic Algorithms (GA), in classifying Algerian SMEs based on their financial health (solvency or insolvency) during the period (2017–2023). The study analyzed financial data from a sample of 100 SMEsregistered in the National Commercial Register (CNRC), utilizing 15 financial ratios (e.g., current ratio, debt ratio, cash ratio) within the Visual Studio Code development environment . The results revealed a significant outperformance of the Genetic Algorithm model optimized with Random Forests, achieving a prediction accuracy of 99.46%, compared to the Deep Neural Network model, which achieved 97.34%. The study also identified specific financial ratios such as the quick liquidity ratio, and return on assets as having the strongest impact on diagnosing financial distress. In contrast, other ratios (e.g., fixed asset turnover ratio) exhibited moderate or weak influence