A prediction histogram is an intuitive graphic way for visualizing model predictions and the separation it achieves for the actual labels.
The models predictions (confidences) are binned and colored according to the true class label. For example, on the figure on the right, red bars are model confidences for negative labeled samples, and green bars are the model confidences for positive labeled samples.
This can help users understand the trade-offs of choosing different thresholds (working points), and find the best threshold according to their need. For example, choosing a threshold of 0.5, will lead to a lot of positive samples which will be classified as negative (i.e false negatives).
A model that can achieve perfect accuracy will show two modalities, where there is never a green bar left of (or together) with a red bar.