Machine Learning Foundations

Intro · 5 concepts · 5 clips · 7 source videos. Watch the clips in order, then take the quick check to see what did not land.

What kind of problem is this

Labelled data and a target to predict is supervised. Structure to find with no target is unsupervised. Almost every modelling mistake starts by getting this wrong.

Steve Brunton (University of Washington) separates supervised from unsupervised learning (2:19-8:07)

Did this clip teach it?

Source videos · 7

Steve Brunton (University of Washington)
Professor of mechanical engineering who teaches machine learning as applied mathematics. Precise about what each method assumes.
Supervised & Unsupervised Machine Learning
Watch full video
DeepLearningAI (Andrew Ng)
Andrew Ng’s course material, posted lecture by lecture. Strongest on the practical diagnosis of a model that is not working.
Bias / Variance (C2W1L02)
Watch full video
StatQuest (Josh Starmer)
The standard reference for statistics and machine learning fundamentals. Works every idea through a small numerical example rather than a formula.
Machine Learning Fundamentals: Cross Validation
Watch full video
StatQuest (Josh Starmer)
The standard reference for statistics and machine learning fundamentals. Works every idea through a small numerical example rather than a formula.
Machine Learning Fundamentals: The Confusion Matrix
Watch full video
StatQuest (Josh Starmer)
The standard reference for statistics and machine learning fundamentals. Works every idea through a small numerical example rather than a formula.
A Gentle Introduction to Machine Learning
Watch full video
StatQuest (Josh Starmer)
The standard reference for statistics and machine learning fundamentals. Works every idea through a small numerical example rather than a formula.
Machine Learning Fundamentals: Bias and Variance
Watch full video
Kimberly Fessel
Data scientist. Short, precise treatments of the metrics people routinely confuse.
Precision vs Recall, With a Clear Example
Watch full video

Ask the sources

Answers are drawn only from the clips on this path. Click a citation to jump the player to that concept.

The ideas in these clips, on Atlas

Precision against recall: which error you can afford is a business decisionDecide which mistake is more expensive before tuning a model; the threshold is a business choice the statistics only express.

Explore Atlas →