Skip to main content
. Author manuscript; available in PMC: 2019 Nov 23.
Published in final edited form as: Metabolomics. 2018 Nov 23;14(12):153. doi: 10.1007/s11306-018-1451-8

Fig. 5.

Fig. 5

Imputation error for kNN and NS-kNN on different versions of MM for the bacterial dataset. Curves represent the mean NRMSE for 100 randomly generated missingness datasets. The percentage of non-high abundance metabolites (I) for all plots ranged from 5 to 70%, as long as the total missingness percentage was not exceeded. II = 70% for all plots. Subpanels show results for a 10% total missingness with III = 30%; b 10% total missingness with III = 40%; c 30% total missingness with III = 30%; and d 30% total missingness with III = 40%. Error bars represent one standard deviation. Asterisks indicate statistical significance (p < 0.05)