Additional Topics in High-Dimensional Data Analysis

Part of the Use R book series (USE R)


In this chapter, we describe several additional approaches that are well suited to high-dimensional data settings. Three of these, namely random forests (RFs) (Section 7.1), logic regression (Section 7.2) and multivariable adaptive regression splines (MARS) (Section 7.3), extend the tree framework outlined in Chapter 6. Random forests were originally proposed by Breiman (2001), and logic regression was first described by Kooperberg et al. (2001), Ruczinski et al. (2003) and Ruczinski et al. (2004). A complete description of the MARS methodology can be found in Friedman (1991).We also present a brief description of Bayesian variable selection (Section 7.4), with particular emphasis on fundamental concepts that will guide the reader in further explorations. Additional readings on Bayesian variable selection methods and related extensions include George and McCulloch (1993), Chipman et al. (1998), Brown et al. (2002), West (2003), Lunn et al. (2006) and Hoggart et al. (2008), among others. Bayesian variable selection approaches are becoming increasingly popular in the analysis of high-throughput genotype data for identifying sets of SNPs that are associated with the trait under investigation. Finally, we end with a listing of several alternative high-dimensional data tools and their sources that may provide additional insight in characterizing genotype{trait associations (Section 7.5).


Random Forest Multivariate Adaptive Regression Spline Additional Topic Importance Score Potential Predictor Variable 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag New York 2009

Authors and Affiliations

  1. 1.University of MassachusettsSchool of Public Health & Health SciencesAmherstUSA

Personalised recommendations