Suchergebnisse

The k-Nearest Neighbour method is a popular nonparametric technique for solving classification and regression problems without having to make potentially restrictive a priori assumptions about the functional form of the statistical relationship under investigation. The purpose of this paper was to demonstrate that the scope of this method can be extended in a way that enables the simultaneous consideration of continuous, ordered discrete, and unordered discrete explanatory variables. An exemplary application to a publicly available dataset demonstrated the feasibility of the proposed approach.