Imputation Of Missing Values In Clinical Data

Micha Johannes BIRKLBAUER

Bakalářská práce

Anotace:

Imputation of missing data is a crucial step in data analysis since many statistical methods require complete datasets. In that regard MissForest imputation is a powerful tool that seems to outperform most other imputation approaches. This analysis evaluates how good imputation using MissForest is compared to other methods like imputation by Multivariate Imputation by Chained Equations (MICE), Restricted Boltzmann Machines (RBM) or the standard strawman (mean) imputation in a clinical dataset that is used to predict the mortality of patients after heart valve surgery.

Abstract:

Imputation of missing data is a crucial step in data analysis since many statistical methods require complete datasets. In that regard MissForest imputation is a powerful tool that seems to outperform most other imputation approaches. This analysis evaluates how good imputation using MissForest is compared to other methods like imputation by Multivariate Imputation by Chained Equations (MICE), Restricted Boltzmann Machines (RBM) or the standard strawman (mean) imputation in a clinical dataset that is used to predict the mortality of patients after heart valve surgery.

Klíčová slova

imputation, missing data, missforest, mice, multivariate imputation by chained equations, rbm, restricted boltzmann machine, clinical data, machine learning

Jazyk práce: čeština

Datum vytvoření / odevzdání či podání práce: 3. 9. 2019

Obhajoba závěrečné práce

• Obhajoba proběhla 9. 9. 2019
• Vedoucí: doc. Dr. Ulrich Bodenhofer

JIHOČESKÁ UNIVERZITA V ČESKÝCH BUDĚJOVICÍCH

Přírodovědecká fakulta

Bakalářský studijní program / obor:
Applied Informatics / Bioinformatics