File Exchange

image thumbnail

Non-crossing polynomial quantile regression

version 1.1.0.0 (3.91 KB) by Wolfgang Schwanghart
Non-crossing polynomial quantile regression

4 Downloads

Updated 17 Jan 2016

GitHub view license on GitHub

ncquantreg finds the coefficients of a polynomial p(x) of degree n that fits the data in vector x to the quantiles tau of y.
ncquantreg(x,y) performs median regression (tau = 0.5) using a polynomial of degree n=1.
ncquantreg(x,y,n,tau) fits numel(tau) polynomials with degree n. The algorithm uses a stepwise multiple quantile regression estimation using non-crossing constraints (Wu and Liu, 2009). The approach is stepwise in a sense that a quantile function is estimated so that it does not cross with a function fitted in a previous step. The algorithm starts from the middle quantile (i.e. the one closest to 0.5) and than progressivly works through the quantiles with increasing distance from the middle.

ncquantreg(x,y,n,tau,pn,pv,...) takes several parameter name value pairs that control the algorithm and plotting.
Reference

Wu, Y., Liu, Y., 2009. Stepwise multiple quantile regression estimation using non-crossing constraints. Statistics and its Interface 2, 299–310.

Comments and Ratings (0)

Updates

1.1.0.0

Changed title

MATLAB Release Compatibility
Created with R2012a
Compatible with any release
Platform Compatibility
Windows macOS Linux
Acknowledgements

Inspired by: quantreg(x,y,tau,order,Nboot)