Theory of Intelligence
MACHINE LEARNING•Supervised Learning•Evaluating Classifiers•2.1 Accuracy & the Confusion Matrix

Introduction

Why accuracy can be dangerously misleading, and how the confusion matrix shows exactly which mistakes a classifier makes.

timer1 min read•format_list_bulletedConcept 1 of 3•forumLesson Discussion

Introduction

Evaluating a classifier is often trickier than evaluating a regression model. With regression, an error of 5,000issimplyanerrorof5,000 is simply an error of 5,000. With classification, not all mistakes are equal:

  • A spam filter that sends an important job offer to the spam folder (false alarm) is annoying.
  • A cancer screening test that tells a sick patient "you're fine" (missed case) can be deadly.

A single number like accuracy treats these the same. This chapter gives you better tools, starting with the most important one: the confusion matrix.

Discuss theoretical concepts, mathematical formulas, and architectural foundations.

Supports inline code with `backticks`
No comments yet. Be the first to start the discussion!
Next Lessonarrow_forward
2.2 Precision, Recall & F1