Nearestneighbor classifiers

Nearest-neighbors classifiers (51), assign samples to classes by matching the gene expression profile to that of samples whose class is known. A simple implementation is to choose a rule for finding the k nearest neighbors and then deciding the classification by majority vote. Nearest-neighbor classifiers are robust, simple to interpret and implement, and do not require, although they may benefit from, preliminary dimension reduction. Nearest-neighbor algorithms are also used in several statistical software packages for imputation of missing data.

0 0

Post a comment