An AUC-based permutation variable importance measure for random forests

Episode von Medizin - Open Access LMU - Teil 21/22 von Ludwig-Maximilians-Universität München · 1. Januar 2013

Background: The random forest (RF) method is a commonly used tool for classification with high dimensional data as well as for ranking candidate predictors based on the so-called random forest variable importance measures (VIMs). However the classification performance of RF is known to be suboptimal in case of strongly unbalanced data, i.e. data where response class sizes differ considerably. Suggestions were made to obtain better classification performance based either on sampling procedures or on cost sensitivity analyses. However to our knowledge the performance of the VIMs has not yet been examined in the case of unbalanced response classes. In this paper we explore the performance of the permutation VIM for unbalanced data settings and introduce an alternative permutation VIM based on the area under the curve (AUC) that is expected to be more robust towards class imbalance. Results: We investigated the performance of the standard permutation VIM and of our novel AUC-based permutation VIM for different class imbalance levels using simulated data and real data. The results suggest that the new AUC-based permutation VIM outperforms the standard permutation VIM for unbalanced data settings while both permutation VIMs have equal performance for balanced data settings. Conclusions: The standard permutation VIM loses its ability to discriminate between associated predictors and predictors not associated with the response for increasing class imbalance. It is outperformed by our new AUC-based permutation VIM for unbalanced data settings, while the performance of both VIMs is very similar in the case of balanced classes. The new AUC-based VIM is implemented in the R package party for the unbiased RF variant based on conditional inference trees. The codes implementing our study are available from the companion website: http://www.ibe.med.uni-muenchen.de/organisation/mitarbeiter/070\_drittmi ttel/janitza/index.html.

Weitere Episoden von „Medizin - Open Access LMU - Teil 21/22“

• A genotype-specific, randomized controlled behavioral intervention to improve the neuroemotional outcome of cardiac surgery: study protocol for a randomized controlled trial 1. Januar 2013
• A new web-based method for automated analysis of muscle histology 1. Januar 2013
• Acceptance of routine or case-based inquiry for intimate partner violence: a mixed method study 1. Januar 2013
• Activation of axonal Kv7 channels in human peripheral nerve by flupirtine but not placebo - therapeutic potential for peripheral neuropathies: results of a randomised controlled trial 1. Januar 2013
• Aequilibrium prudentis: on the necessity for ethics and policy studies in the scientific and technological education of medical professionals 1. Januar 2013
• Autosomal recessive spastic ataxia of Charlevoix Saguenay (ARSACS): expanding the genetic, clinical and imaging spectrum 1. Januar 2013
• `Burden to others' as a public concern in advanced cancer: a comparative survey in seven European countries 1. Januar 2013
• Characterization of in vivo chemoresistant human hepatocellular carcinoma cells with transendothelial differentiation capacities 1. Januar 2013
• Conscious Presence and Self Control as a measure of situational awareness in soldiers - A validation study 1. Januar 2013
• Contrast enhanced transesophageal echocardiography in patients with atrial fibrillation referred to electrical cardioversion improves atrial thrombus detection and may reduce associated thromboembolic events 1. Januar 2013
• Current experience with applying the GRADE approach to public health interventions: an empirical study 1. Januar 2013
• Current management of patients hospitalized with community-acquired pneumonia across Europe: outcomes from REACH 1. Januar 2013
• Diversity of the var gene family of Indonesian Plasmodium falciparum isolates 1. Januar 2013
• Driving forces for changes in geographical distribution of Ixodes ricinus ticks in Europe 1. Januar 2013
• A comparative analysis of the properties of regulated promoter systems commonly used for recombinant gene expression in Escherichia coli 1. Januar 2013

Alle Episoden von Medizin - Open Access LMU - Teil 21/22 · Mehr Bildung-Podcasts