Skip to main content
Fig. 4 | BMC Medical Research Methodology

Fig. 4

From: A comparative study: classification vs. user-based collaborative filtering for clinical prediction

Fig. 4

Simulation results for Chronic Kidney, Dermatology, NHANES, and SUPPORT are shown for a less than 10% missing data, b 10–16% missing data, c 20% missing data, d 30% missing data. Graphs depict the mean estimate of misclassification and standard error calculated via repeated 3-fold cross validation for 50 simulated patterns of missingness for each level of severity. The results for the baseline levels of missingness for each data sets is captured in a and b

Back to article page