You may have already realized that we have discussed ensemble learning. It is defined by www.scholarpedia.org as "the process by which multiple models, such as classifiers or experts, are strategically generated and combined to solve a particular computational intelligence problem". In random forest and gradient boosting, we combined the "votes" of hundreds or thousands of trees to make a prediction. Thus, by definition, those models are ensembles. This methodology can be extended to any learner to create ensembles, which some refer to as meta-ensembles or meta-learners. We will look at one of these methods referred to as "stacking". In this methodology, we will produce a number of classifiers and use their predicted class probabilities as input features to another classifier. This method can result in improved predictive accuracy. In the previous chapters, we focused on classification problems focused on binary outcomes. We will now look at methods to predict those situations where the data consists of more than two outcomes, a very common situation in real-world data sets. I have to confess that the application of these methods in R is some of the most interesting and enjoyable applications I have come across.