Abstract
Financial leverage conditions business stability, although the literature debates whether it depends more on operational efficiency or short-term credit management. The objective is to determine, through machine learning, which indicators most clearly explain the debt ratio. To this end, a Random Forest model was used on a basis of leverage and five efficiency metrics (sales, portfolio, and fixed asset turnover, plus average collection and payment periods). After cleaning the records, the set was divided into training (75%) and testing (25%); the hyperparameters were optimized using random search with cross-validation. Performance was evaluated with R² and RMSE, and the interpretation was supported by permutation importance and SHAP values, complemented by residual plots and relevance networks. The model explained 87% of the variance in training and showed that commercial efficiency, especially portfolio turnover and, secondarily, sales turnover, is the decisive factor in leverage, while collection and payment periods have marginal weight. It is concluded that accelerating cash conversion, rather than modifying credit terms, is the most effective strategy for containing debt

This publication has 13 references indexed in Scilit: