New Zealand Statistical Association 2024 Conference

Yongshi Deng

University of Auckland

Vismi: Visualisation Tools for Multiple Imputation

This is joint work with Thomas Lumley

Multiple imputation has been widely used to handle missing data. There are an increasing number of software packages available for multiple imputation. However, before proceeding with statistical inference, it is crucial for practitioners to assess the quality of multiply-imputed values, particularly when using multiple imputation implementations based on machine learning algorithms. To help with this, we have developed an R package vismi, which offers comprehensive visual diagnostics for evaluating multiple imputation generated by different packages. In this talk, we will demonstrate various functions of vismi, and showcase its practical utility through case studies.