Proizvod vam ne odgovara? Nema veze! Možete nam vratiti unutar 30 dana
S poklon bonom ne možete pogriješiti. Za poklon bon primatelj može odabrati bilo što iz naše ponude.
30 dana za povrat kupljenih proizvoda
Consistency modeling for gene selection is a new topicemerging from recent cancer bioinformatics research. The result ofclassification or clustering on a training set was often found verydifferent from the same operations on a testing set. Here, theissue is addressed as a consistency problem. In practice, theinconsistency of microarray datasets prevents many typical geneselection methods working properly for cancer diagnosis andprognosis. In an attempt to deal with this problem, a new conceptof performance-based consistency is proposed in this thesis. Theproposed consistency concept has been investigated on eightbenchmark microarray and proteomic datasets. The experimentalresults show that the different microarray datasets have differentconsistency characteristics, and that better consistency can leadto an unbiased and reproducible outcome with good diseaseprediction accuracy.