Efficient sparse estimation on interval-censored data with approximated L0 norm: Application to child mortality
Open Access
- 9 April 2021
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 16 (4), e0249359
- https://doi.org/10.1371/journal.pone.0249359
Abstract
A novel penalty for the proportional hazards model under the interval-censored failure time data structure is discussed, with which the subject of variable selection is rarely studied. The penalty comes from an idea to approximate some information criterion, e.g., the BIC or AIC, and the core process is to smooth the ℓ0 norm. Compared with usual regularization methods, the proposed approach is free of heavily time-consuming hyperparameter tuning. The efficiency is further improved by fitting the model and selecting variables in one step. To achieve this, sieve likelihood is introduced, which simultaneously estimates the coefficients and baseline cumulative hazards function. Furthermore, it is shown that the three desired properties for penalties, i.e., continuity, sparsity, and unbiasedness, are all guaranteed. Numerical results show that the proposed sparse estimation method is of great accuracy and efficiency. Finally, the method is used on data of Nigerian children and the key factors that have effects on child mortality are found.Funding Information
- National Natural Science Foundation of China (71571113)
- State Key Program of National Natural Science Foundation of China (91546202)
- Major Science and Technology Projects in Hunan Province (2018GK1020)
- Fundamental Research Funds for the Central Universities
This publication has 27 references indexed in Scilit:
- Nearly unbiased variable selection under minimax concave penaltyThe Annals of Statistics, 2010
- Regression Analysis of Multivariate Interval‐Censored Failure Time Data with Application to Tumorigenicity ExperimentsBiometrical Journal, 2008
- Adaptive Lasso for Cox's proportional hazards modelBiometrika, 2007
- The Adaptive Lasso and Its Oracle PropertiesJournal of the American Statistical Association, 2006
- Regression analysis of interval-censored failure time data with linear transformation modelsThe Canadian Journal of Statistics / La Revue Canadienne de Statistique, 2005
- Sieve Estimation for the Proportional-Odds Failure-Time Regression Model with Interval CensoringJournal of the American Statistical Association, 1997
- Efficient estimation for the proportional hazards model with interval censoringThe Annals of Statistics, 1996
- Semiparametric analysis of the additive risk modelBiometrika, 1994
- Estimating the Dimension of a ModelThe Annals of Statistics, 1978
- A new look at the statistical model identificationIEEE Transactions on Automatic Control, 1974