Figure 8.

In-depth evaluation of RF imputation. Percent bias (accuracy of imputation on raw data) is shown in the upper left for metabolites in bins of percent missingness between 2–51%. Differences in Cronbach’s alpha are shown in the upper left for a variety of proportions of overall missingness between 10–70%. The differences in regression coefficients are shown in the bottom left for a variety of proportions of overall missingness between 10–60%. The respective differences in p-values are shown in the bottom right. The center line represents the median. The lower and upper box limits represent the 25% and 75% quantiles, respectively. Black dots represent single iterations of evaluating differences that are outliers. The whiskers extend to the largest observation within the box limit ± 1.5 × interquartile range.