Spectral embedded generalized mean based k-nearest neighbors clustering with S-distance

作者：

Highlights：

• A churn prediction model is proposed using an enhanced spectral clustering (SC).

• A non-linear distance measure called S-distance is merged with the conventional SC.

• The proposed clustering algorithm is validated on 15 datasets.

• Three state-of-the-art methods are considered to compare with the proposed one.

摘要

•A churn prediction model is proposed using an enhanced spectral clustering (SC).•A non-linear distance measure called S-distance is merged with the conventional SC.•The proposed clustering algorithm is validated on 15 datasets.•Three state-of-the-art methods are considered to compare with the proposed one.

论文关键词：S-distance,Spectral clustering,Symmetry favored k-nearest neighbors,Generalized mean,Distributed computing

论文评审过程：Received 12 March 2020, Revised 13 November 2020, Accepted 14 November 2020, Available online 17 November 2020, Version of Record 10 February 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114326