Hutson AD. Quantile planes without crossing via nonlinear programming.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE 2018;
153:185-190. [PMID:
29157450 PMCID:
PMC10091397 DOI:
10.1016/j.cmpb.2017.10.019]
[Citation(s) in RCA: 2] [Impact Index Per Article: 0.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Received: 03/21/2017] [Revised: 09/18/2017] [Accepted: 10/13/2017] [Indexed: 06/07/2023]
Abstract
BACKGROUND AND OBJECTIVE
In this note we propose a nonlinear programming approach for simultaneous fitting of quantile regression models for two or more quantiles. The approach is straightforward, flexible and practical. We apply this approach to a dataset of lactic acid values from a screening dataset in childhood malaria.
METHODS
We carry out the fitting of simultaneous quantile regression models using a specific definition of a quantile as an expectation via nonlinear programming methods given certain monotonicity constraints.
RESULTS
We illustrate through simulations and examples that are new approach to simultaneous quantile regression is practical and feasible. The approach is supplemented by providing a bootstrap framework for confidence interval estimation.
CONCLUSIONS
Our nonlinear programming approach towards solving the simultaneous quantile regression fitting is shown to be a practical approach that should appeal to statistical practitioners.
Collapse