Parameter-free surrounding neighborhood based regression methods

doi:10.1016/j.eswa.2022.116881

Parameter-free surrounding neighborhood based regression methods

2022-08-01

1

PRE

AI

T

Tülin İnkaya *

DOI：10.1016/j.eswa.2022.116881

Original

Original request for help

Save

Abstract

En 中文

In machine learning, nearest neighbor (NN) regression is one of the most prominent methods for numericprediction. It estimates the output variable of a new data point by averaging the output variables of theneighboring points. The selection of the neighborhood and its parameter(s) is crucial for the performanceof NN regression, however this is still an open issue. This study contributes to the literature by adoptingthe parameter-free surrounding neighborhood (PSN) concept for NN regression. PSNs are based on proximitygraphs, i.e. minimum spanning tree, relative neighborhood graph, and Gabriel graph. They yield a uniqueneighborhood for each point by combining proximity, connectivity and spatial distribution. The performancesof the PSN regression methods are compared with k-nearest neighbors, k-nearest centroid neighbors, andsupport vector regression using real-world data sets. The statistical tests show that the PSN regression methodsperform significantly better than most of the competing approaches. Also, the proposed approaches do not haveany parameters to be set.

Keywords:

Prediction

k-nearest regression

Minimum spanning tree

Relative neighborhood graph

Gabriel graph

Parameter-free surrounding neighborhood based regression methods

Abstract

Journal

Organization