arXiv 2007.070522020-07-14stat.MEcs.LG

在没有真值的情况下预测特征归因性

Predicting feature imputability in the absence of ground truth

Ulster University（阿尔斯特大学）；National University of Ireland Galway（爱尔兰国立大学戈尔韦分校）；Altnagelvin Area Hospital（阿尔特纳格尔文地区医院）；Western Health and Social Care Trust（西部健康与社会保健信托）

Niamh McCombe, Xuemei Ding, Girijesh Prasad, David P. Finn, Stephen Todd, Paula L. McClean, KongFatt Wong-Lin

备注：: 5 pages, 3 figures, 1 table. In: Proceedings of the 37th International Conference on Machine Learning (ICML), 2020

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

摘要：

数据插补是处理缺失值的最常用方法，但在大多数实际应用中，可能会出现大量缺失数据，并且很难或不可能评估数据是否已准确插补（缺乏真值）。本文针对这些问题，提出了一个有效的和简单的主成分为基础的方法，以确定是否可以准确地填补个别数据功能-功能可归责性。特别是，我们建立了一个强大的主成分负荷和功能归咎之间的线性关系，即使在极端缺失和缺乏真值的存在。这项工作将有重要的影响，在实际的数据估算策略。

英文摘要：

Data imputation is the most popular method of dealing with missing values, but in most real life applications, large missing data can occur and it is difficult or impossible to evaluate whether data has been imputed accurately (lack of ground truth). This paper addresses these issues by proposing an effective and simple principal component based method for determining whether individual data features can be accurately imputed - feature imputability. In particular, we establish a strong linear relationship between principal component loadings and feature imputability, even in the presence of extreme missingness and lack of ground truth. This work will have important implications in practical data imputation strategies.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。