Contribution to the research of Anthropometric measurements and comparison of body proportions in the student population in Bangladesh

This study presents a novel approach to developing an adaptive clothing size chart and assessing wearability for men aged 20–25. Anthropometric data were collected and analyzed using hierarchical clustering and k-means to identify distinct body measurement groups. Key parameters, including height, weight, hip circumference, and shoulder width, were used to classify individuals into four distinct groups, forming the basis for a size chart reflecting actual body measurements. Additionally, quantitative deviations between garment dimensions and body measurements were evaluated. The overall fit loss for the population was calculated at 1010.16, with significant deviations observed in hip and shoulder width measurements, highlighting the need for dynamic sizing adjustments. An improved regression model utilizing gradient boosting was employed to predict key anthropometric measurements. The model demonstrated superior performance, achieving an R 2 score of 0.647 where the mean squared error value is 18.62 , surpassing both linear regression and random forest models. A data-driven framework was established to develop flexible size charts for pants (waist, hips, thighs, inseam, outer seam) and shirts (bust, shoulder, back, waist, hips), accommodating body variability through group-based dimensions and dynamic ranges. This integration of statistical and machine learning methods enhances garment fit, reduces fit loss, and improves consumer satisfaction. These findings provide valuable insights for the apparel industry, facilitating the development of adjustable fabric designs and optimized sizing strategies. • This study collected anthropometric data from 497 men for size chart development. Following ISO 8559-1 and ASTM D5585-10 standards, 29 body measurements were selected. The selection of the 29 body measurements in your study aligns with guidelines from ISO 8559–1:2017 and ASTM D5585-11 standards, which provide comprehensive frameworks for anthropometric measurements in clothing design. • After constructing the size chart for basic shirts and pants, fit loss was calculated to refine the sizing model. • The gradient boosting model demonstrated the best performance for the Bangladeshi male population, achieving an R 2 score of 0.647, where the mean squared error value is 18.62. • This study collected anthropometric data from 497 men for size chart development. Following ISO 8559-1 and ASTM D5585-10 standards, 29 body measurements were selected. The selection of the 29 body measurements in your study aligns with guidelines from ISO 8559–1:2017 and ASTM D5585-11 standards, which provide comprehensive frameworks for anthropometric measurements in clothing design. • After constructing the size chart for basic shirts and pants, fit loss was calculated to refine the sizing model. • The gradient boosting model demonstrated the best performance for the Bangladeshi male population, achieving an R 2 score of 0.647, where the mean squared error value is 18.62.