Better Quantile Contour for Constructing Growth Charts

Abstract

In this paper we going to comparison using Mean Absolute Error (MAE), among three nonparametric methods: Nadaraya-Watson (NW), B-Splines (BS) and Smoothing Spline (SS) which use to estimate quantile functions depending on the model of time-varying coefficient under the concept of depth in the bivariate case, to prepare for construct a better contours.

We used simulation study for construct two stages: first stage is to construct the simulation models for the three nonparametric methods and then comparison among them by using MAE, where the method that gives the least MAE is the best method, and the second stage is to plot the best method which given the best contour shape.

Where it was concluded that the BS is the best method because it's given a suitable contouring shape and less MAE which equal to 0.048 which will be used to plot best bivariate growth charts.