Statistika: Statistics and Economy Journal (Jun 2016)

Segmented Regression Based on Cut-off Polynomials

  • Miloš Kaňka

Journal volume & issue
Vol. 96, no. 2
pp. 60 – 72

Abstract

Read online

In Statistika: Statistics and Economy Journal No. 4/2015 (pp. 39–58), author’s paper Segmented Regression Based on B-splines with Solved Examples was published. Use of B-spline basis functions has many advantages, the most important being a special form of matrix of system of normal equations suitable for quick solution of this system. The subject of this paper is to explain how that segmented regression can be mathematically developer in other way, which doesn’t require the knowledge of relatively complicated theory of B-spline basis functions, but is based on simpler apparatus of cut-off polynomials. The author considers a detailed calculation of matrix of system of normal equations elements and elaboration of so called polygonal method, as his contribution to issues of segmented regression. This method can be used to automatically obtain required values of nodal points. Author pays major attention to computing elements of matrix of systém of normal equations, which he also developed as computer program called TRIO.

Keywords