About Support Vector Machines

Link to Jupyter Notebook

This tutorial is adapted from Jake VanderPlas’s example of SVM as given in his reference book: Python Data Science Handbook

Motivation for Support Vector Machines

We want to find a line/curve (in 2D) or a manifold (in n-D) that divides the class from each other. This is a type of Discriminative Classification

Consider a simple case of classification task, in which the two classes of points are well separated. We can find region in space which best separates the data into two classes. The Support Vectors in Support Vector Machine are the (hyper)planes which lie at the edge of the individual classes. This idea is much easier to understand from 2D perspective.

Nifty tech tag lists fromĀ Wouter Beeftink