{"title":"夏普利值的因果分析:条件值与边际值","authors":"Ilya Rozenfeld","doi":"arxiv-2409.06157","DOIUrl":null,"url":null,"abstract":"Shapley values, a game theoretic concept, has been one of the most popular\ntools for explaining Machine Learning (ML) models in recent years.\nUnfortunately, the two most common approaches, conditional and marginal, to\ncalculating Shapley values can lead to different results along with some\nundesirable side effects when features are correlated. This in turn has led to\nthe situation in the literature where contradictory recommendations regarding\nchoice of an approach are provided by different authors. In this paper we aim\nto resolve this controversy through the use of causal arguments. We show that\nthe differences arise from the implicit assumptions that are made within each\nmethod to deal with missing causal information. We also demonstrate that the\nconditional approach is fundamentally unsound from a causal perspective. This,\ntogether with previous work in [1], leads to the conclusion that the marginal\napproach should be preferred over the conditional one.","PeriodicalId":501425,"journal":{"name":"arXiv - STAT - Methodology","volume":"192 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-09-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Causal Analysis of Shapley Values: Conditional vs. Marginal\",\"authors\":\"Ilya Rozenfeld\",\"doi\":\"arxiv-2409.06157\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Shapley values, a game theoretic concept, has been one of the most popular\\ntools for explaining Machine Learning (ML) models in recent years.\\nUnfortunately, the two most common approaches, conditional and marginal, to\\ncalculating Shapley values can lead to different results along with some\\nundesirable side effects when features are correlated. This in turn has led to\\nthe situation in the literature where contradictory recommendations regarding\\nchoice of an approach are provided by different authors. In this paper we aim\\nto resolve this controversy through the use of causal arguments. We show that\\nthe differences arise from the implicit assumptions that are made within each\\nmethod to deal with missing causal information. We also demonstrate that the\\nconditional approach is fundamentally unsound from a causal perspective. This,\\ntogether with previous work in [1], leads to the conclusion that the marginal\\napproach should be preferred over the conditional one.\",\"PeriodicalId\":501425,\"journal\":{\"name\":\"arXiv - STAT - Methodology\",\"volume\":\"192 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-10\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - STAT - Methodology\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.06157\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - STAT - Methodology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.06157","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Causal Analysis of Shapley Values: Conditional vs. Marginal
Shapley values, a game theoretic concept, has been one of the most popular
tools for explaining Machine Learning (ML) models in recent years.
Unfortunately, the two most common approaches, conditional and marginal, to
calculating Shapley values can lead to different results along with some
undesirable side effects when features are correlated. This in turn has led to
the situation in the literature where contradictory recommendations regarding
choice of an approach are provided by different authors. In this paper we aim
to resolve this controversy through the use of causal arguments. We show that
the differences arise from the implicit assumptions that are made within each
method to deal with missing causal information. We also demonstrate that the
conditional approach is fundamentally unsound from a causal perspective. This,
together with previous work in [1], leads to the conclusion that the marginal
approach should be preferred over the conditional one.