#What it does
It finds the optimal boundary (hyperplane) with the maximum margin that separates data points from different classes (i.e. classifier), ensuring robust generalization to unseen data.
#How it works
- Identifies the hyperplane that best separates data points of different classes.
- For non-linearly separable data, SVM uses a kernel trick, i.e., mathematical functions that transform data into higher dimensions, e.g. Polynomial Kernel or Radials Basis Function (RBF)
#Preconditions
- Scaled data
- Minimal noise
#Evaluation
#Advantages
- Effective in high dimensions
- Robust to overfitting
- Versatile
#Limitations
- Computationally complex
- Requires tuning
- Difficult to interpret