K-Nearest Neighbours
The K-nearest neighbours stores all available cases and classifies new cases by a majority vote of its k-neighbors. The case being assigned to the class is most common amongst its K nearest neighbours measured by a distance function. This means that the algorithm takes a bunch of labelled points, and learns from them how to label other points by looking at the closest points to the new point (i.e. its nearest neighbours) and having the neighbours vote – so that the label most commonly selected for the new vote is the final label for the new point.
Applications
Predict gene function from heterogeneous data ref.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.