SL is a generalization of the stacking algorithm [32], an ensemble machine learning method that takes a weighted average of all other algorithms considered for prediction and produces a single prediction function (PF) with optimal tradeoff between variance and bias.