Skip to main content
Fig. 5 | BMC Medical Research Methodology

Fig. 5

From: Treatment of missing data in Bayesian network structure learning: an application to linked biomedical and social survey data

Fig. 5

Performance on MCAR data with 1000 data points. Precision (A) and recall (B) of three different methods of handling incomplete data: none, multiple imputation by chained equations (MICE) and structural expectation-maximization (SEM). Rows represent different missing proportions and columns indicate different groups of number of variables. Barplots show means with error bars representing standard error of the mean. Adjusted p-values for ANOVAs are displayed in those panels that are significant at least the 0.05 level. Lines representing significant Tukey’s HSD pairwise tests are shown and annotated as: *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001

Back to article page