Bias in random forest variable importance measures: Illustrations, sources and a solution

Episode von Medizin - Open Access LMU - Teil 15/22 von Ludwig-Maximilians-Universität München · 1. Januar 2007

Background: Variable importance measures for random forests have been receiving increased attention as a means of variable selection in many classification tasks in bioinformatics and related scientific fields, for instance to select a subset of genetic markers relevant for the prediction of a certain disease. We show that random forest variable importance measures are a sensible means for variable selection in many applications, but are not reliable in situations where potential predictor variables vary in their scale of measurement or their number of categories. This is particularly important in genomics and computational biology, where predictors often include variables of different types, for example when predictors include both sequence data and continuous variables such as folding energy, or when amino acid sequence data show different numbers of categories. Results: Simulation studies are presented illustrating that, when random forest variable importance measures are used with data of varying types, the results are misleading because suboptimal predictor variables may be artificially preferred in variable selection. The two mechanisms underlying this deficiency are biased variable selection in the individual classification trees used to build the random forest on one hand, and effects induced by bootstrap sampling with replacement on the other hand. Conclusion: We propose to employ an alternative implementation of random forests, that provides unbiased variable selection in the individual classification trees. When this method is applied using subsampling without replacement, the resulting variable importance measures can be used reliably for variable selection even in situations where the potential predictor variables vary in their scale of measurement or their number of categories. The usage of both random forest algorithms and their variable importance measures in the R system for statistical computing is illustrated and documented thoroughly in an application re-analyzing data from a study on RNA editing. Therefore the suggested method can be applied straightforwardly by scientists in bioinformatics research.

Weitere Episoden von „Medizin - Open Access LMU - Teil 15/22“

• Occupation and adult onset of rhinitis in the general population 1. Januar 2008
• Retention of mouth-to-mouth, mouth-to-mask and mouth-to-face shield ventilation 1. Januar 2008
• An in vivo evaluation of Brilliant Blue G in animals and humans 1. Januar 2008
• Syk-mediated translocation of PI3Kδ to the leading edge controls lamellipodium formation and migration of leukocytes 7. November 2007
• Evolutionary aspects in evaluating mutations in the melanocortin 4 receptor 1. Oktober 2007
• Natural killer cells promote early CD8 T cell responses against cytomegalovirus. 24. August 2007
• Analysis of intraviral protein-protein interactions of the SARS coronavirus ORFeome 23. Mai 2007
• The association of a SNP upstream of INSIG2 with body mass index is reproduced in several but not all cohorts 1. April 2007
• Absence of Ret signaling in mice causes progressive and late degeneration of the nigrostriatal system 1. März 2007
• Antithrombin and hypercoagulability in sepsis: insights from thrombelastography? 1. Januar 2007
• Evidence for placebo effects on physical but not on biochemical outcome parameters: a review of clinical trials 1. Januar 2007
• Effect of the angiotensin receptor blocker irbesartan on metabolic parameters in clinical practice: the DO-IT prospective observational study 1. Januar 2007
• Acute childhood diarrhoea in northern Ghana: epidemiological, clinical and microbiological characteristics 1. Januar 2007
• Gene transfer of wild-type apoA-I and apoA-I Milano reduce atherosclerosis to a similar extent 1. Januar 2007
• Does respiratory health contribute to the effects of long-term air pollution exposure on cardiovascular mortality? 1. Januar 2007

Alle Episoden von Medizin - Open Access LMU - Teil 15/22 · Mehr Bildung-Podcasts