Performance comparison of feature selection and extraction methods with random instance selection
作者：
Highlights：
• We evaluate a framework for reducing run time of big data feature reduction methods.
• This framework selects a small random subset of instances prior to feature reduction.
• We present comprehensive computational experiments using large public datasets.
• Execution time can be reduced by a factor of 90 with minimal impact on performance.
• We provide recommendations on which feature reduction method to use with a classifier.
摘要
•We evaluate a framework for reducing run time of big data feature reduction methods.•This framework selects a small random subset of instances prior to feature reduction.•We present comprehensive computational experiments using large public datasets.•Execution time can be reduced by a factor of 90 with minimal impact on performance.•We provide recommendations on which feature reduction method to use with a classifier.
论文关键词：Explainable artificial intelligence,Dimension reduction,Feature selection,Feature extraction,Instance selection,Data preprocessing
论文评审过程：Received 10 February 2020, Revised 10 December 2020, Accepted 15 April 2021, Available online 23 April 2021, Version of Record 5 May 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115072
