Machine Learning in Elixir

Learning from Your Surroundings

Linear models are one of many traditional machine learning models implemented in Scholar. Another common model you’ll come across in the wild is the K-Nearest Neighbors (KNN) model. The idea behind KNN models is relatively simple: you are likely very similar to the things nearest to you. Intuitively this makes sense. If you were to pluck a random person off the streets of Philadelphia and ask them their favorite football team, they’d likely say the Eagles. That’s because the Eagles are the hometown team, and they’re close in proximity to where the Eagles play. As you drift away from Philadelphia, the distribution of favorite NFL teams changes.

KNN works simply by classifying new points as belonging to the most ...