An optimisation approach to apparel sizing
- PDF / 227,626 Bytes
- 8 Pages / 595 x 842 pts (A4) Page_size
- 83 Downloads / 230 Views
#1998 Operational Research Society Ltd. All rights reserved. 0160-5682/98 $12.00 http://www.stockton-press.co.uk/jor
An optimisation approach to apparel sizing CE McCulloch, B Paal and SP Ashdown Cornell University, USA A novel approach for the construction of apparel sizing systems is formulated. As a ®rst step to this process, ef®cient sizing systems are de®ned based on a mathematical model of garment ®t. Nonlinear optimisation techniques are then used to derive a set of possible sizing systems using multidimensional information from anthropometric data. The method is illustrated by developing a sizing system designed for a dress shirt of a military uniform using anthropometric data from the US Army. Results of this analysis show that endogenous size assignment and selection of disaccommodated individuals, together with relaxation of the requirement of a `stepwise' size structure, results in substantial improvements in ®t over an existing sizing system. The proposed methodology enables the development of sizing systems that can either increase accommodation of the population, reduce the number of sizes in the system, or improve overall ®t in accommodated individuals. Keywords: anthropometry; apparel ®t; decision theory; Nelder±Mead simplex method
Introduction In the most general terms, the purpose of an apparel sizing system is to divide a varied population into homogeneous subgroups. Members of a subgroup are similar to each other in body size and shape so that a single garment can adequately ®t each of them. Members of different subgroups are dissimilar and would therefore require different garments. The primary sizing system used by the US apparel industry for women's clothing relies on a pattern grading process that sets ®xed proportional increments between sizes. The body dimensions used to create this sizing system originated from an outdated anthropometric study of civilian women conducted by O'Brien and Shelton.1 The sizes that make up the system were derived from regression relationships between selected body dimensions from this study. Over time, individual companies modi®ed their sizes based on anecdotal information generated from consumer feedback. However, Hudson2 and McVey3 report that most companies do not have systematic methods for developing sizes or for determining what range of consumers are being ®tted by their sizing system. Somewhat more systematic methods and current anthropometric data are used by the armed forces to create sizing systems. Regression analysis is used to identify the two key body dimensions that are considered the most essential to garment ®t and function, and that are also highly correlated Correspondence: Dr SP Ashdown, Department of Textiles and Apparel, Cornell University, Ithaca, NY 14853, USA.
with other dimensions important for the garment being sized. The values of these other dimensions are derived from multiple regressions using the appropriate key body dimension as the independent variable.4 However, the actual sizes within the system are set in a heuristic ma
Data Loading...
