Is the Classification: Predicting a Category lesson free?

Yes. The first 5 lessons on PyRun are free forever — no signup, no card. Classification: Predicting a Category runs real Python in your browser via Pyodide.

How long does the Classification: Predicting a Category lesson take?

About 14 minutes end-to-end. It's a hands-on lesson — you read a paragraph, edit the code, and run it in the same tab.

Do I need to install Python to try the Classification: Predicting a Category lesson?

No install required. PyRun runs real CPython in your browser via Pyodide (Python compiled to WebAssembly). Works on Chrome, Firefox, Safari, and Edge — desktop or mobile.

What level is the Classification: Predicting a Category lesson?

Advanced level. Assumes comfort with classes, functions, and standard library. Not for first-time Python learners.

← All lessons

Advanced·14 min·machine-learning · sklearn · classification

Classification: Predicting a Category

When the answer is a category — pass/fail, spam/not-spam, which of 3 products — that's classification. The workhorse starting model is logistic regression (a classifier, despite the name).

Same two verbs, different output

fit and predict work exactly as before. The difference: predict returns a class (0 or 1) instead of a number. And you get a bonus method:

predict_proba(X) — the model's confidence for each class. 0.92 means "92% sure this student passes." Probabilities let you set your own threshold instead of blindly trusting the default 0.5 — critical when a false negative costs more than a false positive.

Accuracy — and its trap

accuracy_score = fraction predicted correctly. Simple, but misleading on imbalanced data: if 95% of emails are "not spam," a model that always says "not spam" scores 95% while catching zero spam. When classes are lopsided, accuracy alone lies — you'll need precision/recall (next lessons) to see the truth.

Try it

Print clf.predict_proba for a [[1, 3]] student. Is the model confident they fail?
Flip one label in y and re-fit. How much does test accuracy move? Small datasets are fragile.

Practice

3 graded exercises

Write the code, click Check. We’ll tell you exactly what to fix.

Progress0 / 3

Exercise 1
Write fit_classifier(X, y) returning a fitted LogisticRegression.
Exercise 2
Write pass_probability(model, X_row) returning the probability of class 1 for one row X_row (a list of features).
Exercise 3
Write test_accuracy(X, y) that splits (30% test, random_state=0), fits logistic regression on train, and returns test accuracy.

← PreviousRegression: Predicting a Number Next →Decision Trees & Feature Importance

Classification: Predicting a Category

Same two verbs, different output

fit and predict work exactly as before. The difference: predict returns a class (0 or 1) instead of a number. And you get a bonus method:

predict_proba(X) — the model's confidence for each class. 0.92 means "92% sure this student passes." Probabilities let you set your own threshold instead of blindly trusting the default 0.5 — critical when a false negative costs more than a false positive.

Accuracy — and its trap

Try it

Print clf.predict_proba for a [[1, 3]] student. Is the model confident they fail?
Flip one label in y and re-fit. How much does test accuracy move? Small datasets are fragile.

Practice

3 graded exercises

Write the code, click Check. We’ll tell you exactly what to fix.

Progress0 / 3

Exercise 1
Write fit_classifier(X, y) returning a fitted LogisticRegression.
Exercise 2
Write pass_probability(model, X_row) returning the probability of class 1 for one row X_row (a list of features).
Exercise 3
Write test_accuracy(X, y) that splits (30% test, random_state=0), fits logistic regression on train, and returns test accuracy.