A simulation study on missing data imputation for dichotomous variables using statistical and machine learning methods

Study design

The framework of the study design was shown in Fig. 1, which consisted of four main steps: generating specific missing scenarios by simulation, data imputation, performance evaluation, and statistical test.

Generating specific missing scenarios by simulation

The pseudocode in Table 1 shows the basic flow of the simulation study. The simulation study considered multiple factors including missing mechanisms, sample sizes, missing rates, the correlation between variables, value distributions, and the number of missing variables. Rubin DB proposed three missing mechanisms…