Investigating Association Between BMI and Other Variables Using Skew-Symmetric Regression Models

Abstract

Finding significant predictors of body mass index (BMI) drew researcher attention for decades due to the well-known fact that very high BMI leads to obesity [1]. Most of the previous researchers used ordinary least squares (OLS) regression for this purpose in which the error term follows normal distribution [2]. Since the distribution of BMI is skewed [3], ordinary regression may not be suitable to determine significant covariates of BMI. In this project, we used two real life datasets and used multiple skew-symmetric regression model to identify variables that affects BMI. We also compared our results obtained from the skew-symmetric models to the results from OLS regression.