Machine Learning · Chapter 15 of 40
Support Vector Machines (SVM)
SVMs find the optimal boundary that separates classes by the widest MARGIN.
With kernels (RBF, polynomial), they can capture complex non-linear boundaries.
Example 1 (python)
from sklearn.svm import SVC
m = SVC(kernel='rbf', C=1.0).fit(X_train, y_train)
print(m.score(X_test, y_test))Output
0.88RBF kernel handles non-linear boundaries.
Example 2 (python)
# Always scale features for SVMsScale-sensitive like KNN.
Key points
- Finds max-margin boundary.
- Kernels enable non-linear boundaries.
- Requires scaled features.
- Slower on very large datasets.
💡 Note: C controls how much the model tolerates misclassifications. Small C = wider margin, more errors allowed.