Classification trees

Classification trees recursively partition the space of expression profiles into subsets that are highly predictive of the phenotype of interest (38). They are robust, easy-to-use, and can automatically sift large data sets, identifying important patterns and relationships. No prescreening of the genes is required. The resulting predictive models can be displayed using intuitive graphical representations. An example in which classification trees have been applied to gene expression data can be found in Zhang and Yu (60).

0 0

Post a comment