A new approach to estimate anthropometric measurements by adaptive neuro-fuzzy inference system

Abstract

Abstract Eighteen anthropometric measurements were taken in standing and sitting positions, from 387 subjects between 15 and 17 years old. “Adaptive Neuro-Fuzzy Inference System (ANFIS)” was used to estimate anthropometric measurements as an alternative to stepwise regression analysis. Six outputs (shoulder width, hip width, knee height, buttock-popliteal height, popliteal height, and height) were selected for estimation purpose. The results showed that the number of inputs required estimating outputs varied with sex difference. ANFIS perform better than stepwise regression method for both sex groups, as revealed by the standard deviations averaged over the six outputs: S ANFIS =0.776, S Regression =0.855 for boys and, S ANFIS =0.883, S Regression =1.027 for girls. Relevance to industry More recently, as industry and marketing reach around the globe, body size has become a matter of practical interest to designers and engineers. In ergonomics anthropometric data are widely used to specify the physical dimension of workspaces, equipment, furniture and clothing. This is especially true for school children, which spend most of their time sitting at their chairs and desks and ought to be able to adopt comfortable body postures.