The main objective of this research is to "improve production performance" in the spinning and single-ply yarn production section in the textile industry by leveraging artificial intelligence (AI) and appropriate machine learning algorithms, while identifying an optimal algorithm for advancing intelligent manufacturing in textile production. Production data were analyzed using regression-based machine learning models to enhance yarn production performance in the worsted spinning system, minimize machine downtime and material waste, and ultimately improve overall production efficiency. The results demonstrate that machine learning models can effectively predict key production performance indicators in the textile industry, enabling reductions in material waste and unplanned downtime and thereby enhancing production efficiency during the single-ply yarn manufacturing stage on ring spinning machines at Guilan Spinning Company. Among the evaluated models, the Gradient Boosting Regressor achieved the highest predictive accuracy and overall performance, with an R² score of 0.938 and an RMSE of 22.182, making it the most effective model for data analysis and production performance prediction. Compared with the other models, the Gradient Boosting Regressor offers several advantages, including a lower risk of overfitting, the ability to efficiently process structured datasets with minimal preprocessing, high predictive accuracy, faster computational performance, and shorter training time. Furthermore, SHAP (SHapley Additive exPlanations) analysis revealed that the most influential factors affecting production performance, ranked by their contribution to the model's predictions, are yarn count, primary cause of machine stoppage, amount of material waste, downtime duration, and description of the primary cause of machine stoppage.