Skip to content
Back to formatted view

Raw Message

Message-ID: <2475A5E7ACB80F4C81BFF348ADCC9B0F4E635EDD@EXMBX03.xuhb.nhs.uk>
Date: 2016-04-26T12:25:19Z
From: David Mcnulty
Subject: Penalised spline regression

Good Afternoon Everyone,

I am looking for advice fitting a linear mixed model where the random components do not seem to fit within the model formulae for lmer. The columns of Z are not stratified and have the notional random formula (z1 | 1) + ... + (zk | 1). 

Context
I am fitting a penalised thin plate spline with knots k1 to kn. The basis functions Zk are |x-ki|^3 and the penalty matrix has elements |ki-kj|^3. Rather than fit basis functions Zk I have transformed the basis to Z = Zk %*% penalty^(-1/2).

Full details are in section 2: Bayesian analysis for penalised spline regression using winBUGS. https://www.jstatsoft.org/article/view/v014i14/v14i14.pdf

I am specifically looking for an alternative to the function spm() in package semipar.

Thank you for any help or advice.

Dave.

David Mcnulty
Medical Statistician
 
 
Tel: +44 (0) 121 371 2448  
Internal: 12448     
Email: David.Mcnulty at uhb.nhs.uk 
Web: http://www.uhb.nhs.uk  


Health Informatics - University Hospitals Birmingham NHS Foundation Trust
Yardley Court, 11-13 Frederick Road, Edgbaston
Birmingham, B15 1JD


The Health Informatics Department is a team of specialist NHS information analysts who work together to provide analysis solutions. The Team filters, analyses and presents comprehensive information that empowers NHS professionals to make informed decisions.