Chapter 7. The General Ensemble Technique
The previous four chapters have dealt with the ensembling techniques for decision trees. In each of the topics discussed in those chapters, the base learner was a decision tree and, consequently, we delved into the homogenous ensembling technique. In this chapter, we will demonstrate that the base learner can be any statistical or machine learning technique and their ensemble will lead to improved precision in predictions. An important requirement will be that the base learner should be better than a random guess. Through R programs, we will discuss and clarify the different possible cases in which ensembling will work. Voting is an important trait of the classifiers – we will state two different methods for this and illustrate them in the context of bagging and random forest ensemblers. The averaging technique is an ensembler for regression variables, which will follow the discussion of classification methods. The chapter will...
