Please enable JavaScript.

Coggle requires JavaScript to display documents.

Chpt 5- Overfitting (Overfitting (Examined section (Learning curves (X=How…

- - - - Trees are flexible
    - - Adding attributes, especially non-linear (X^2) makes a function more complex
    - - Support vector machine
        
        Can cut groups when outliers are present
    - - Complexity yields correlations unable to generalize to pop
    - - Split non-holdout into groups
        
        Model created for each k-1
        
        Test that against remaining group
    - - X=How much data
      - Y=Accuracy
- - - - 2 methods
        
        Choose amount via hypothesis testing
        
        Pruning method