The UU-test for statistical modeling of unimodal data
作者：
Highlights：
• We propose UU-test that is a new method for deciding on dataset unimodality.
• The method attempts to approximate the empirical cdf of the dataset using a cdf that is piecewise linear, unimodal and models the data sufficiently.
• A unique feature is that the method also provides a statistical model of a unimodal dataset in the form of a uniform mixture model.
• The method can be employed in various data analysis tasks (e.g. clustering) where deciding on dataset unimodality is important.
摘要
•We propose UU-test that is a new method for deciding on dataset unimodality.•The method attempts to approximate the empirical cdf of the dataset using a cdf that is piecewise linear, unimodal and models the data sufficiently.•A unique feature is that the method also provides a statistical model of a unimodal dataset in the form of a uniform mixture model.•The method can be employed in various data analysis tasks (e.g. clustering) where deciding on dataset unimodality is important.
论文关键词：Unimodal data,Unimodality test,Statistical modeling,Uniform mixture model
论文评审过程：Received 21 July 2020, Revised 17 August 2021, Accepted 20 August 2021, Available online 29 August 2021, Version of Record 4 September 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2021.108272
