AdaBoost.SAMME.R is a variant that works with classifiers that can output prediction probabilities. This is normally possible employing techniques such as Platt scaling, but it's important to check whether a specific classifier implementation is able to output the probabilities without any further action. For example, SVM implementations provided by Scikit-Learn don't compute the probabilities unless the parameter probability=True (because they require an extra step that could be useless in some cases).
In this case, we assume that the output of each classifier is a probability vector:
Each component is the conditional probability ...
