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

Alphonce Bere; Godfrey H. Sithuba; Coster Mabvuu; Retang Mashabela; Caston Sigauke; Kwabena Kyei

doi:10.37920/sasj.2021.55.1.3

Alphonce Bere Department of Statistics, University of Venda, Thohoyandou, South Africa
Godfrey H. Sithuba Department of Statistics, University of Venda, Thohoyandou, South Africa
Coster Mabvuu Department of Statistics, University of Venda, Thohoyandou, South Africa
Retang Mashabela Department of Statistics, University of Venda, Thohoyandou, South Africa
Caston Sigauke Department of Statistics, University of Venda, Thohoyandou, South Africa
Kwabena Kyei Department of Statistics, University of Venda, Thohoyandou, South Africa

DOI: https://doi.org/10.37920/sasj.2021.55.1.3

Keywords: Boosting, Discrete survival model, First alcohol intake, Penalized likelihood, Variable selection

Abstract

We present the results of a simulation study performed to compare the accuracy of a lassotype 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.

Downloads

Download data is not yet available.