Fig. 1
From: The effect of noise on the predictive limit of QSAR models
Graphical representation of experimental error and prediction error for an arbitrary dataset. Each row of bars is a separate observation for an arbitrary endpoint. The red bars represent population means (“true” values) for the observation, the grey bars represent experimental values, and the blue bars represent prediction values. If prediction values are closer to the population means than the experimental values, the true error will be smaller than the observed error