Abstract
In quantile regression, various quantiles of a response variable Y are modelled as func- tions of covariates (rather than its mean). An important application is the construction of reference curves/surfaces and conditional prediction intervals for Y. Recently, a nonparametric quantile regression method based on the concept of optimal quantization was proposed. This method competes very well with k-nearest neighbor, kernel, and spline methods. In this paper, we describe an R package, called QuantifQuantile, that allows to perform quantization-based quantile regression. We describe the various functions of the package and provide examples.
Highlights
In numerous applications, quantile regression is used to evaluate the impact of a d-dimensional covariate X on a response variable Y
We described the package QuantifQuantile that allows to implement the quantizationbased quantile regression method introduced in Charlier et al (2015a,b)
Since the choice of the tuning parameter N is crucial, a warning message is printed if it is not well-chosen and the function plot can be used as guide to change adequately the value of the parameter testN in the various functions
Summary
Quantile regression is used to evaluate the impact of a d-dimensional covariate X on a (scalar) response variable Y. The quantile functions x → qα(x) provide reference curves (when d = 1), one for each value of α. For fixed x, they provide conditional prediction intervals of the form Iα = [qα(x), q1−α(x)] (α < 1/2). They provide conditional prediction intervals of the form Iα = [qα(x), q1−α(x)] (α < 1/2) Such reference curves and prediction intervals are widely used, e.g. in economics, ecology, or lifetime analysis. In medicine, they are used to provide reference growth curves for children’s height and weight given their age
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
