Predictive modeling of tearing strength in laser-engraved denim garments using Multiple Linear Regression

Abstract

This study presents the development of a Multiple Linear Regression (MLR) model to predict the tearing strength of laser-engraved denim garments in both the warp and weft directions, based on key input parameters: Dot Per Inch (DPI) and pixel time. The model achieved excellent predictive accuracy, with R² values of 0.9967 for the warp direction and 0.9911 for the weft direction, indicating that over 99% of the variability in tearing strength was explained by the model. The Pearson correlation coefficients (0.9983 for warp, 0.9956 for weft) and Spearman’s rank correlation coefficients (1 for warp, 0.9833 for weft) further confirm the strength of the relationship between the predicted and actual values. The Mean Absolute Percentage Error (MAEP) values of 0.8783% (warp) and 1.6837% (weft) demonstrate the model’s high accuracy, with significantly lower errors compared to previous fuzzy logic model. Residual analysis confirmed the assumptions of normality, homoscedasticity, and independence, validating the model’s reliability. The MLR model provides a robust tool for optimizing laser engraving parameters in denim manufacturing, reducing the need for trial-and-error testing and ensuring consistent product quality.