Creating R System for Optimization of Real-Time Physiologic Parameters: Application of Loess Regression

• Activity Number: 187
• Type: Topic Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Sports
• Author(s): Joshua Callaway*+ and D. Keith Williams and Zoran Bursac and Benjamin Stone
• Companies: University of Arkansas for Medical Sciences and University of Arkansas for Medical Sciences and University of Arkansas for Medical Sciences and Sigma Human Performance
• Keywords: Loess Regression ; R ; Nonparametric Regression ; Nonlinear Regression ; VO2 Max

Abstract

Customized models inform individuals on parameter interpretation during exercise in order to maximize fat loss. Due to R's free open-source nature, we have augmented established statistical techniques into a unique system available on demand for parameter optimization. Loess regression serves as an excellent tool for this undertaking due to its flexibility. It takes the best aspects from both regression worlds of linear least squares and nonlinear regression. Loess operates by fitting local models to smaller pieces of the entire dataset, thus adapting to changes in variation throughout the regression. Therefore, we have implemented the R loess command into a function that can plot one or two response variables against the desired predictor variable for individual datasets. In essence, an individual wishing to optimize fat burning can "plug in" variables and pinpoint where to maintain intensity during exercise.