Dimension Reduction of Anthropometric Measurements with Support Vector Machine for Regression: Application to a French Military Personnel Database

Abstract

Collecting anthropometric data is a heavy and time-consuming procedure. The aim of this study was to find a reduced set of anthropometric measurements able to estimate the full-body dimensions of a given individual. The method was developed and applied on a database of 122 measurements carried out on 459 females and 771 males of the French military personnel. Among the 122, 26 key measurements were chosen. A regression method based on support vector machine was used to predict these key measurements in relation to each other. The designed “minimal measurement set selecting algorithm” chose 6 main inputs to predict the remained 20 measurements with mean correlation of 0.94 and 0.92, respectively on the training and on the testing data. This result tends to prove that the regression method can be used to predict the French military personnel anthropometrics.