Machine Learning: Theory and Applications · University of Alberta
Every model in this course, as something you can move
The material of EDPY 506, one topic at a time. A lecture describes what a model does. Here you set the parameter yourself and watch what it does to the fit, on data the page generates while you read. Nothing goes beyond what the course covered.
How to read this
Each topic is one page, and each page is built the same way: the idea in words, then a figure you can interact with, then what the figure shows once you have played with it. The figures are not illustrations of the maths. They run the maths, in your browser, on data the page generates while you watch.
Every page uses the same example, the Edmonton home prices above, so a model on one page can be compared with a model on another without re-learning a dataset. The homes are simulated rather than real, so the right answer is known, and a model that has recovered it can be told apart from one that has only appeared to. The example follows the one used in class. The figures are built from scratch rather than adapted from the demos shown in the lectures.