Supervised learning: classifying and predicting
The approach shared by every supervised model: features and target, classification or regression, training then predicting, decision boundary, data leakage and what each of the course's algorithms assumes.
- 25 cards
- 6 topics
- Free
The outline, topic by topic
Each topic is weighted the way the exam weighs it.
- Concepts
- Questions
- Sort and link
- The numbers
- The curves
- Scenarios
A few cards from the deck
Supervised learning: what is it?
Show the answer
Learning from examples whose expected answer is known, in order to predict that answer on new cases.
Every training example carries its label. Without labels (grouping customers without knowing in advance which ones are alike), it is unsupervised learning.
The survival of Titanic passengers is predicted from the columns of a real dataset. Which one must never be given to the model?
boat, the lifeboat numberpclass, the class of the passenger's ticketsex, the sex recorded for the passengerage, the passenger's age on board the ship
Show the answer
boat, the lifeboat numberHaving a lifeboat number almost means having survived: that column is filled in after the event, it contains the answer. Class, sex and age are known before the sinking, they are genuine explanatory features.
To train a supervised model, it is given a matrix of … written X and a vector of … written y.
Show the answer
To train a supervised model, it is given a matrix of features written X and a vector of targets written y.
Each row of X describes one individual and the matching row of y gives the expected answer. After training, the model receives only X.