Constructing Multivariate Survival Trees: The MST Package for R
Constructing Multivariate Survival Trees: The MST Package for R
复制标题
DOI：
10.18637/jss.v083.i12
复制
发表时间：
2018-02-01
影响因子：
5.8
通讯作者：
Fan, Juanjuan
中科院分区：
文献类型：
--
作者：
Calhoun, Peter;Su, Xiaogang;Fan, Juanjuan
多元生存树需要很少的统计假设，易于解释，并提供有意义的诊断和预测规则。树可以处理大量具有混合类型的预测器，并且不需要预测器变量转换或选择。这些在许多应用领域都是有用的功能，并且在当前的大数据时代经常需要。本文的目的是介绍R包MST。该软件包使用基于边缘模型和脆弱模型的方法构建多变量生存树。它允许用户控制和查看树是如何构建的。该软件包还可以模拟来自边缘和脆弱模型的高维多变量生存数据。
Multivariate survival trees require few statistical assumptions, are easy to interpret, and provide meaningful diagnosis and prediction rules. Trees can handle a large number of predictors with mixed types and do not require predictor variable transformation or selection. These are useful features in many application fields and are often required in the current era of big data. The aim of this article is to introduce the R package MST. This package constructs multivariate survival trees using marginal model and frailty model based approaches. It allows the user to control and see how the trees are constructed. The package can also simulate high-dimensional, multivariate survival data from marginal and frailty models.