4.7. Ensemble Learners

In supervised data mining, the objective is to build a model that can explain the relationship between inputs and output. The model can be considered a hypothesis that can map new input data to predicted output. For a given training set, multiple hypotheses can explain the relationship with varying degrees of accuracy. While it is difficult to find the exact hypothesis from an infinite hypothesis space, we would like the modeling process to find the hypothesis that can best explain the relationship with least error.
Ensemble methods or learners optimize the hypothesis-finding problem by employing an array of individual prediction models and then combining them to form an aggregate hypothesis or model. These methods provide ...

Get Predictive Analytics and Data Mining now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.