Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Random forest-based imputation outperforms other methods for imputing LC-MS metabolomics data: a comparative study

Fig. 4

Summary Results. Summary results for six imputation methods; MIN, MEAN, BPCA, PPCA, KNN and RF. The general trend of the methods is being presented here with the y-axis being the average NRMSE for the four percentages of missing values (5, 10, 20 and 30%) together after 100 permutations. Each line shows one missing mechanisms (represented by a different color); MCAR, MAR, MNAR, MCAR-MAR, MCAR-MNAR, MAR-MNAR, MCAR-MAR-MNAR, and each black dot represents the average NRMSE with the error bars being the standard deviations of the NRMSEs for 100 permutations. The error bars are useful here because they report the uncertainty of the estimation of the imputed value per method

Back to article page