Fig. 3
From: 1H NMR based urinary metabolites profiling dataset of canine mammary tumors

Performance evaluation of missing value imputation methods. Four imputation methods were evaluated along with the percentage of missing numbers (x-axis). (a) Normalized root mean squared error ranking plot. (b) Procrustes sum of squared error plot. QRILC, quantile regression imputation of left-censored data; HM, half minimum; GSimp, an iterative Gibbs sampler based left-censored missing value imputation approach.