I’m not a great machine learning guy, but I’m pretty happy with how it all turned out. Kanber’s explanations are approachable and doesn’t get too bogged down in theory. These are all pretty introductory, but still useful. And, it’s nice seeing some Machine learning logic in a language I’m already comfortable with. Admittedly, I’m not sure all the articles strictly fall under the umbrella of “Machine Learning”, like Full-text search, but it’s still relevant.
- k-nearest neighbor: Which pre-existing category does a new data point belong to
- k-means clustering: Categorizes data into one of ‘k’ clusters
- genetic algorithms: “Mates” and mutates data to find local optimum
- naive bayes: Handy for document classification (spam vs ham, or language it’s written )
- full text search: How to rank documents when searching for keywords across them
These gripes are relatively minor. Kanber explains the concepts well, and I’m grateful he published these articles and open sourced all of his code. It’s been a couple years since the last article, but hopefully he’ll deliver more.