runTwoLayerExtCV-methods {Rmagpie} | R Documentation |
This method run an external two-layers cross-validation according to the options stored in an object of class assessment. The concept of two-layers cross-validation has been introduced by J.X. Zhu,G.J. McLachlan, L. Ben-Tovim Jonesa, I.A.Wood in 'On selection biases with prediction rules formed from gene expression data' and by I. A. Wood, P. M. Visscher, and K. L. Mengersen in 'Classification based upon gene expression data: bias and precision of error rates' (cf. section References). This technique of cross-validation is used to determine an unbiased estimate of the best error rate (using the best size of subset for RFE-SVM, of the best threshold for NSC) when feature selection is involved.
object |
Object of class assessment . Object assessment of interest |
object of class assessment
in which the one-layer external cross-validation
has been computed, therfore, the slot resultRepeated2LayerCV
is no more NULL.
This methods print out the key results of the assessment, to access the full detail
of the results, the user must call the method getResults
.
J.X. Zhu, G.J. McLachlan, L. Ben-Tovim, I.A. Wood (2008), "On selection biases with prediction rules formed from gene expression data", Journal of Statistical Planning and Inference, 38:374-386.
I.A. Wood, P.M. Visscher, and K.L. Mengersen "Classification based upon gene expression data: bias and precision of error rates" Bioinformatics, June 1, 2007; 23(11): 1363 - 1370.
assessment
, getResults
, runOneLayerExtCV-methods
data('vV70genesDataset') # assessment with RFE and SVM myExpe <- new("assessment", dataset=vV70genes, noFolds1stLayer=9, noFolds2ndLayer=10, classifierName="svm", typeFoldCreation="original", svmKernel="linear", noOfRepeat=2, featureSelectionOptions=new("geneSubsets", optionValues=c(1,2,3,4,5,6))) myExpe <- runTwoLayerExtCV(myExpe)