Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Mixture density networks for the indirect estimation of reference intervals

Fig. 4

Average prediction error (independent setting). Average prediction error over the full range of the input variable for the parameters of the first component (top three rows) and the 95% quantile (bottom row). Predictions are based on the results without location crossing from the simulation setup with mixture weights independent from the input and sample size \(n=10000\)

Back to article page