SVM Decision Boundary Visualizer

Visualize how support vector machines create decision boundaries for classification

Overview

The SVM Decision Boundary Visualizer lets you explore how Support Vector Machines create decision boundaries to classify data. Experiment with different kernels (linear, polynomial, RBF, sigmoid) and parameters to see how they affect the boundary shape and model performance.

Tips

1. Start with linearly separable data using the linear kernel to understand the basics before moving to complex patterns
2. Try the XOR preset with an RBF kernel to see how non-linear kernels handle data that can’t be separated by a straight line
3. Adjust the C parameter to observe the trade-off between margin width and allowing misclassifications (low C = wider margin, high C = stricter boundary)
4. Experiment with gamma on RBF kernels - low values create smooth boundaries, high values create complex, tightly-fitted boundaries
5. Watch the support vectors (highlighted points) - notice that only these critical points define the decision boundary, not all training data