Evaluating the Cyclical Hybrid Imputation Technique (CHIT) under MCAR, MAR, and MNAR missing data mechanisms: Evidence from health datasets
Kurban Kotan, Serdar Kırışoğlu, Hiroyuki Maruyama · 2026
Read Evaluating the Cyclical Hybrid Imputation Technique (CHIT) under MCAR, MAR, and MNAR missing data mechanisms: Evidence from health datasets by Kurban Kotan, Serdar Kırışoğlu, Hiroyuki Maruyama (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13623145.txt.
Description
Kurban Kotan, Serdar Kırışoğlu, and Hiroyuki Maruyama test a technique for filling gaps in health data, a step called imputation. Missing values plague medical records, and different filling methods fail in different ways depending on why data went missing. Their technique, named CHIT, was compared with two methods under three missing-data patterns.
The team worked with three health data sets, on kidney disease, heart disease, and mouse protein measures, deleting a fifth to a quarter of values in planned patterns. Eight classification programs were trained on the filled data. CHIT held near-perfect accuracy across patterns, while rival methods lost ground in the hardest pattern, where missingness depends on the unseen values. The authors credit two design choices: the training pool grows as records are completed, and the scarcest fields per record are filled first. They offer the results as guidance for data teams.
Catalog record
| CREATOR | Kurban Kotan, Serdar Kırışoğlu, Hiroyuki Maruyama |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |