Regularisation in discrete survival models: A comparison of lasso and gradient boosting

IF 0.4 Q4 STATISTICS & PROBABILITY

SOUTH AFRICAN STATISTICAL JOURNAL Pub Date : 2021-03-31 DOI:10.37920/SASJ.2021.55.1.3

A. Bere, Godfrey H. Sithuba, Coster Mabvuu, Retang Mashabela, C. Sigauke, K. Kyei

引用次数: 0

Abstract

We present the results of a simulation study performed to compare the accuracy of a lasso-type penalization method and gradient boosting in estimating the baseline hazard function and covariate parameters in discrete survival models. The mean square error results reveal that the lasso-type algorithm performs better in recovering the baseline hazard and covariate parameters. In particular, gradient boosting underestimates the sizes of the parameters and also has a high false positive rate. Similar results are obtained in an application to real-life data.

查看原文本刊更多论文

离散生存模型中的正则化:套索和梯度提升的比较

我们提出了一项模拟研究的结果，以比较套索类型惩罚方法和梯度增强在估计离散生存模型中基线风险函数和协变量参数时的准确性。均方误差结果表明，套索算法在恢复基线风险和协变量参数方面具有较好的效果。特别是，梯度增强低估了参数的大小，也有很高的假阳性率。在对实际数据的应用中也得到了类似的结果。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

SOUTH AFRICAN STATISTICAL JOURNAL STATISTICS & PROBABILITY-

CiteScore

0.30

自引率

0.00%

发文量

期刊介绍： The journal will publish innovative contributions to the theory and application of statistics. Authoritative review articles on topics of general interest which are not readily accessible in a coherent form, will be also be considered for publication. Articles on applications or of a general nature will be published in separate sections and an author should indicate which of these sections an article is intended for. An applications article should normally consist of the analysis of actual data and need not necessarily contain new theory. The data should be made available with the article but need not necessarily be part of it.