K-Means and Hierarchical clustering both assume our groups are roughly globular. But what about ring-shaped clusters? Crescent moons? Data scattered w...
Parathan ThiyagalingamK-Means demanded that we know K up front. Today's algorithm is the opposite. It builds a tree of every possible clustering, from "everything is one bi...
Parathan ThiyagalingamSeventeen days of supervised learning, where every example came with an answer attached. Today we cross into a different world. The model has to find ...
Parathan ThiyagalingamOur last supervised algorithm. Before deep learning conquered everything in the 2010s, SVMs were the king of classification. They are still genuinely ...
Parathan ThiyagalingamRandom Forest trained 500 trees independently and averaged them. Today we meet a smarter idea. Train trees one at a time, with each new tree focused o...
Parathan ThiyagalingamA single decision tree is fragile. Today we fix that with one of the most famous ideas in ML: instead of trusting one tree, trust a thousand. Random F...
Parathan ThiyagalingamIf we have ever played the game Twenty Questions, we have already done decision trees in our head. Today we automate that intuition. Decision Trees ar...
Parathan ThiyagalingamYesterday's KNN looked at neighbours. Today's algorithm takes a completely different route: pure probability. Naive Bayes is the algorithm that quietl...
Parathan ThiyagalingamEleven days of preparation and we are finally meeting our first classification algorithm beyond Logistic Regression. KNN (K-Nearest Neighbours) is the...
Parathan Thiyagalingam