Machine learning prediction model for the seam strength of five pocket denim jeans for lock stitch

This study develops a CatBoost-based computational model to predict the seam strength of five-pocket denim jeans, a key factor in garment durability. It explores the effects of thread count, stitch density (SPI), fabric weight, and seam types on seam strength. Findings show that thread count has the most significant impact, followed by SPI and fabric weight. Lapped seams exhibit higher strength compared to superimposed seams with the same parameters. The CatBoost model outperforms traditional polynomial regression, achieving a mean relative absolute error of 2.27 % and 1.17 % for superimposed and lapped seams, respectively, with maximum errors of 5.87 % and 2.51 %. The model explains over 95 % of data variance, confirmed by K-fold cross-validation. This tool helps manufacturers optimize sewing parameters, reduce material waste, and improve denim quality. The research highlights the efficacy of advanced machine learning algorithms like CatBoost in enhancing predictive accuracy and manufacturing processes.