{"title":"Statistical Inference for Variable Importance","authors":"M. J. van der Laan","doi":"10.2202/1557-4679.1008","DOIUrl":null,"url":null,"abstract":"Many statistical problems involve the learning of an importance/effect of a variable for predicting an outcome of interest based on observing a sample of $n$ independent and identically distributed observations on a list of input variables and an outcome. For example, though prediction/machine learning is, in principle, concerned with learning the optimal unknown mapping from input variables to an outcome from the data, the typical reported output is a list of importance measures for each input variable. The approach in prediction has been to learn the unknown optimal predictor from the data and derive, for each of the input variables, the variable importance from the obtained fit. In this article we propose a new approach which involves for each variable separately 1) defining variable importance as a real valued parameter, 2) deriving the efficient influence curve and thereby optimal estimating function for this parameter in the assumed (possibly nonparametric) model, and 3) develop a corresponding double robust locally efficient estimator of this variable importance, obtained by substituting for the nuisance parameters in the optimal estimating function data adaptive estimators. We illustrate this methodology in the context of prediction, and obtain in this manner double robust locally optimal estimators of marginal variable importance, accompanied with p-values and confidence intervals. In addition, we present a model based and machine learning approach to estimate covariate-adjusted variable importance. Finally, we generalize this methodology to variable importance parameters for time-dependent variables.","PeriodicalId":50333,"journal":{"name":"International Journal of Biostatistics","volume":"2 1","pages":""},"PeriodicalIF":1.2000,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.2202/1557-4679.1008","citationCount":"163","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"International Journal of Biostatistics","FirstCategoryId":"100","ListUrlMain":"https://doi.org/10.2202/1557-4679.1008","RegionNum":4,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 163
Abstract
Many statistical problems involve the learning of an importance/effect of a variable for predicting an outcome of interest based on observing a sample of $n$ independent and identically distributed observations on a list of input variables and an outcome. For example, though prediction/machine learning is, in principle, concerned with learning the optimal unknown mapping from input variables to an outcome from the data, the typical reported output is a list of importance measures for each input variable. The approach in prediction has been to learn the unknown optimal predictor from the data and derive, for each of the input variables, the variable importance from the obtained fit. In this article we propose a new approach which involves for each variable separately 1) defining variable importance as a real valued parameter, 2) deriving the efficient influence curve and thereby optimal estimating function for this parameter in the assumed (possibly nonparametric) model, and 3) develop a corresponding double robust locally efficient estimator of this variable importance, obtained by substituting for the nuisance parameters in the optimal estimating function data adaptive estimators. We illustrate this methodology in the context of prediction, and obtain in this manner double robust locally optimal estimators of marginal variable importance, accompanied with p-values and confidence intervals. In addition, we present a model based and machine learning approach to estimate covariate-adjusted variable importance. Finally, we generalize this methodology to variable importance parameters for time-dependent variables.
期刊介绍:
The International Journal of Biostatistics (IJB) seeks to publish new biostatistical models and methods, new statistical theory, as well as original applications of statistical methods, for important practical problems arising from the biological, medical, public health, and agricultural sciences with an emphasis on semiparametric methods. Given many alternatives to publish exist within biostatistics, IJB offers a place to publish for research in biostatistics focusing on modern methods, often based on machine-learning and other data-adaptive methodologies, as well as providing a unique reading experience that compels the author to be explicit about the statistical inference problem addressed by the paper. IJB is intended that the journal cover the entire range of biostatistics, from theoretical advances to relevant and sensible translations of a practical problem into a statistical framework. Electronic publication also allows for data and software code to be appended, and opens the door for reproducible research allowing readers to easily replicate analyses described in a paper. Both original research and review articles will be warmly received, as will articles applying sound statistical methods to practical problems.