Abstract
The generalised maximum likelihood estimator (GMLE) of a survival function (Formula presented.) based on truncated interval-censored (TIC) data has been studied since 1990s (by Frydman, H. (1994), ‘A note on nonparametric estimation of the distribution function from interval censored and truncated data’, Journal of the Royal Statistical Society, Series B, 56, 71–74 among others). In the literature related to the GMLE based on TIC data, there are several issues that have not been properly settled in both methodology and theory including: (1) innermost intervals based on the TIC data are not correctly formulated and they lead to inconsistent estimators which are not the GMLE; and (2) the consistency of the GMLE has not been established. We settle these two issues in this paper. In particular, we specify the correct forms of innermost intervals and establish consistency results for the GMLE under a realistic model.
| Original language | English |
|---|---|
| Pages (from-to) | 266-282 |
| Number of pages | 17 |
| Journal | Journal of Nonparametric Statistics |
| Volume | 35 |
| Issue number | 2 |
| DOIs | |
| State | Published - 2023 |
Keywords
- Kullback–Leibler inequality
- Nonparametric maximum likelihood estimator
- consistency
- self-consistent algorithm
- truncated interval censorship model
Fingerprint
Dive into the research topics of 'The generalised MLE with truncated interval-censored data'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver