Please enable JavaScript.

Coggle requires JavaScript to display documents.

Data Mining Algorithms - Coggle Diagram

- - - - Recursive splitting.
      - Uses perpendicular decision boundaries.
      - Strength: Handles categorical data well.
  - - - Creates linear decision boundaries (lines, planes, hyperplanes)
      - Fits weights to maximize classification accuracy.
  - - - Key Concepts:
        
        Maximizes margin.
        
        Hinge loss penalizes misclassified points.
      - Linear SVM: Works with linear separable data.
      - Non-linear SVM: Uses kernel functions to map features to higher dimensions.
  - - - Outputs probabilities of class membership.
      - Log-odds transformation.
      - Common in fraud detection and binary classification tasks.
- - - - Influence the importance of features.
      - Large weights indicate critical features.
      - Weights close to zero suggest less relevant features.
- - - - Limited to linear boundaries but computationally simpler.
      - Logistic regression, SVMs, linear discriminant functions.
    - - Non-linear SVMs with kernels.
        
        Neural networks as "stacked models" for complex relationships.
        
        Example: Adding squared terms creates curved decision boundaries.