THE APPLICATION OF PCR AND PLSR IN MODELING OF BODY FAT PERCENTAGE ESTIMATES
Hedi
Politeknik Negeri Bandung
Jl. Geger kalong Hilir, Ciwaruga
Bandung 40559, Indonesia
Abstract
The body fat content if can be predicted high or low, can be used as a reference to see the level of health. The higher the fat content, the greater the risk of various diseases. In this paper will propose modeling fat content that depends on 13 input variable are age, weight, height, neck, chest, abdomen, hip, thigh, knee, ankle, biceps, forearm, and wrist, The problem faced in this model is that between variables are correlated (multicollinearity). The objective of this paper is determine a fit model for predicted body fat percentage estimates.To overcome these problems applied method of principal component regression (PCR) and partial least square regression (PLSR). By applying the akaike-s information criterion (AIC), mean square error (MSE) and mean absolute percentage error (MAPE), the PLSR model was the method of choice for modeling body fat percentage estimates.
Keywords: Body fat, Modeling, Multicollinearity, PCR, PLSR
Topic: Basic Science in Engineering Education