DataMiningandKnowledgeDiscovery(2021)35:1393–1434 https://doi.org/10.1007/s10618-020-00731-7 Extendinggreedyfeatureselectionalgorithmstomultiple solutions Giorgos Borboudakis 1 ·Ioannis Tsamardinos 1,2,3 Received:31March2019/Accepted:16December2020/Publishedonline:1May2021 ©TheAuthor(s)2021 Abstract Most feature selection methods identify only a single solution. This is acceptable for predictivepurposes,butisnotsufficientforknowledgediscoveryifmultiplesolutions exist. We propose a strategy toextend a class of greedy methods toefficiently identify multiple solutions, and show under which conditions it identifies all solutions. We also introduce a taxonomy of features that takes the existence of multiple solutions intoaccount.Furthermore,weexploredifferentdefinitionsofstatisticalequivalenceo

Join for free to read