Ensemble Learning

In ensemble learning, multiple weak learners is used and combine their prediction by some weighted voting to get the final prediction.

Frequently used weak learner is Decision Tree

Three main types of ensemble learning methods are:

  1. Boosting
  2. Bagging
  3. Stacking or Meta Model in Ensemble Learning
  • The reason for ensemble learning works because the models are UNCORRELATED
    • Main point here is the models and/or data need to be uncorrelated, meaning they (model and/or data) should be different in nature
    • I.E. several version of SVMs with different Hyperparameters may not result better

Pasted image 20231107182645.png