0

Is there a straightforward procedure for converting/representing a fitted B-spline to standard polynomial equations for each ‘between-knot’ piece/segment?

It seems it should be possible but I can’t find a straightforward description of how spline basis coefficients and knot positions relate to coefficients of the corresponding (piecewise) polynomials. Intuitively it would seem likely to be simple operation but I don’t have sufficient depth of knowledge to be confident in deriving it myself.

I’ve seen that there is a Matlab function (fn2fm) for changing between ‘B-spline basis form’ and ‘piecewise polynomial form’ and I’d like to do something similar in R.

rgm
  • 11
  • 2
  • Is this a continuation of your [previous post](https://stackoverflow.com/questions/59414275/inferring-expressing-the-polynomial-equation-of-a-fitted-smoothing-spline)? There was a comment there referring you to the `rms` package of Frank Harrell? Does that not work for you? – user2474226 Dec 23 '19 at 17:57
  • Thanks for the comment; I was trying to generalise the question in the hope of getting a response. Although the rms implementation is useful to know about it uses a specific type of spline and I'm interested in _how_ to convert from one form to the other. – rgm Dec 23 '19 at 19:04

0 Answers0