Application of Data Mining Techniques for Establishing Standard Sizing System for Sudanese Army Officers Garments

Abstract

The purpose of this research was to use data mining techniques in establishing standard sizing system for Sudanese army officers garment based on an anthropometric body measurements variables. It also looked into the role of classification of the whole body types (upper and lower), due to its importance for garments manufacturers. From Sudanese army officers’ database, a data was selected for establishing sizing system for producing army officers uniform (Poshirt) and to propose a new standard sizing system. The anthropometric data was collected from Sur military clothing factory in Sudan. The data set collected for 841 army officers was used .For each individual (13) anthropometric variables were involved. In this research data mining methods (WEKA and SPSS) were used forclustering and establishing sizing system. The K-means algorithm was implemented to determine the final cluster classification in two stages in the first stage, the (WEKA) clustering method was applied on the original database to categories the size figures. The (SPSS) method, as quantitative approach, was used in the second stage in order to generate descriptive statistics for the raw data following a step by step procedure. Cluster analysis using chest and waist as a control anthropometric variables revealed eight distinct clusters namely; XS, S, M, L, XL, XXL, XXXL and XXXXL respectively. Size codes and upper and lower size limits are generated. The new proposed sizing system SUD profile was compared with profile of three national standards sizing charts USA, EUR, and SURThe results revealed that the proposed sizing system follow approximately the same profile as the three national standards sizes charts chosen.