Types of ML Problems

All three are used very commonly, and often in conjunction with each other!

Classification

Given an input, predict what class it is in (discrete). Examples:

  • Given a picture of a handwritten letter, predict what letter it actually is

  • Given the previous 10 days of weather, predict how the sky will be (clear, cloudy, partly cloudy, etc)

  • Given a chess board, predict which move a player should make next

Regression

Given an input, predict a numerical value (continuous). Examples:

  • Estimating stock prices

  • Estimating the tomorrow's temperature

  • Estimating how long it will take you to drive somewhere

Clustering

Given unlabeled data, cluster them into natural groups. Examples:

  • Identifying close friend groups on Facebook

  • Generating common groups of people on Spotify based on their music tastes

Practice

Based on the above, what do you think Netflix show recommendations use? It's not obvious, nor is it just one of them :)

PreviousMachine Learning: Common AlgorithmsNextK-Nearest Neighbors

Last updated