Model‐Based Recursive Partitioning for Discrete Event Times.
Saved in:
| Title: | Model‐Based Recursive Partitioning for Discrete Event Times. |
|---|---|
| Authors: | Huber, Cynthia1 (AUTHOR) cynthia.huber@med.uni-goettingen.de, Schmid, Matthias2 (AUTHOR), Friede, Tim1 (AUTHOR) |
| Source: | Statistical Analysis & Data Mining. Apr2025, Vol. 18 Issue 2, p1-13. 13p. |
| Subjects: | Recursive partitioning, Permutations, Survival rate, Regression analysis, Unemployment, Error probability |
| Abstract: | Model‐based recursive partitioning (MOB) is used to identify subgroups based on various outcome measures, including time‐to‐event outcomes. Discrete time‐to‐event data are typically fitted using the generalized linear model (GLM) framework with binary outcome. However, direct application of MOB with GLMs for binary outcomes needs an augmented data matrix, which violates the assumption of independent observations required for MOB's splitting criterion. We propose a permutation approach tailored to discrete time‐to‐event data that controls the error rate of MOB's splitting criterion, avoiding spurious subgroup identification. In simulations, we compare this permutation approach to the naïve approach applying MOB with regression models for binary outcomes directly to the augmented data and MOB using the sum of the score contributions. Our experiments showed that MOB using the permutation approach controls the Type I error rate and accurately identifies splitting variables. To illustrate the various MOB approaches for discrete time‐to‐event data, we apply them to an example data set on unemployment duration. [ABSTRACT FROM AUTHOR] |
| Copyright of Statistical Analysis & Data Mining is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Model‐based recursive partitioning (MOB) is used to identify subgroups based on various outcome measures, including time‐to‐event outcomes. Discrete time‐to‐event data are typically fitted using the generalized linear model (GLM) framework with binary outcome. However, direct application of MOB with GLMs for binary outcomes needs an augmented data matrix, which violates the assumption of independent observations required for MOB's splitting criterion. We propose a permutation approach tailored to discrete time‐to‐event data that controls the error rate of MOB's splitting criterion, avoiding spurious subgroup identification. In simulations, we compare this permutation approach to the naïve approach applying MOB with regression models for binary outcomes directly to the augmented data and MOB using the sum of the score contributions. Our experiments showed that MOB using the permutation approach controls the Type I error rate and accurately identifies splitting variables. To illustrate the various MOB approaches for discrete time‐to‐event data, we apply them to an example data set on unemployment duration. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 19321864 |
| DOI: | 10.1002/sam.70020 |
