Lesson 0001

From data to decisions

The one diagram that everything else in this course hangs off. If you remember only one thing from Lecture 3, remember this.

What "learning" means

STOR 323 defines machine learning operationally: performance on a task T improves through experience E, measured by performance P. For MNIST digit recognition: T = recognize a handwritten digit, E = labeled examples, P = error on new digits. Nothing mystical — just data in, better performance out.

Two phases, two different objects

The whole course splits into two phases that are easy to blur together if you don't separate them explicitly. Training happens once, ahead of time. Deployment happens over and over, on new inputs.

D --A--> D

Training: a fixed learning algorithm A reads a dataset D and returns a fitted model f̂D.

x --f̂D--> s --δ--> a

Deployment: a new input x is predicted into a score s, then decided by a rule δ into an action a.

The load-bearing distinction: A (the algorithm) is a fixed procedure — it doesn't change. f̂D (the fitted model) is data-dependent state — it changes whenever D changes. δ (the decision rule) and a (the action) belong to the deployed system, not to the model itself. Three different things can each change independently, and STOR 323 will keep asking you to say which one moved.

Why this matters later

Every later topic in this deck is really a question about one arrow in this diagram:

QuestionWhich arrow
Which functions could f̂D possibly be? (linear regression, softmax regression, kNN)D --A--> f̂_D — the model family constrains A
How good is f̂D's prediction s? (MSE, ROC/AUC, confusion matrix)x --f̂_D--> s
What action should a score trigger, and who bears the cost of a wrong one?s --δ--> a

The deck's punchline (slide 21): the model is fixed; the policy is not. You can swap δ — change a threshold, change who gets referred for a test — without retraining anything. That's a deployment-time decision, not a modeling one. Future lessons will dig into model families and evaluation metrics separately; this lesson is just about being able to place any concept on this diagram without hesitation.

Check your recall

Answer from memory before checking — that effort is what makes it stick.

In D --A--> f̂_D, what do we call the object f̂_D itself?
What do we call the δ that maps a score s to an action a?
What do we call the set of candidate functions chosen before any data is seen?

Go deeper

Primary source for this lecture's formalism: An Introduction to Statistical Learning (James, Witten, Hastie, Tibshirani) — free official PDF, Ch. 2 covers exactly this train/predict framing. See RESOURCES.md for more.

Stuck on anything above? Just ask your teaching agent — it can explain any piece of this differently.