Beyond trees and ensembles, a few classic algorithms remain essential tools. Each one takes a very different approach to classification, and knowing them helps you pick the right model for a problem.
You'll start with k-nearest neighbors, which classifies a point by looking at the points closest to it. Next comes the support vector machine, which finds the boundary that separates classes with the widest margin. Finally you'll learn Naive Bayes, a fast probabilistic classifier, including its Gaussian version for numeric data.
Watch the videos in order, then try all three on the same dataset and compare how they perform. When you're done, continue on to Feature Engineering & Pipelines.
How k-NN classifies a new point by voting among its closest neighbors, and how to choose the value of k.
How a support vector machine picks the maximum margin boundary, and how soft margins handle overlapping classes.
How Naive Bayes uses Bayes theorem and the independence assumption to classify text and other categorical data.
How Naive Bayes handles numeric features by modeling each one with a normal distribution.