K-Nearest Neighbor Learning

• most basic method - all instances are points in the n-dimensional space
• distance is defined as standard Euclidean distance, so an instance is where is the rth attribute on instance x.
• The distance between two instances and is

  

• the target value may either be discrete or real-valued
• for the discrete case where V is the finite set and the algorithm returns the most common value of f among the k training examples.
• hypothesis space described by the Voronoi diagram
• for continuous target,

  

  then