Supervised Learning

Supervised Learning is a paradigm in Machine Learning, where we train the computer to predict an output based on the input data. The dataset contains both features and targets, provided by a human.

Notation

Notation	Description
$S$	Dataset. Contains all the features and targets for our predictions
$N$	Total number of samples in the dataset
$x$	Feature. Input value(s) from the dataset. Also called: explicative variables, or class in classification
$X$	Matrix of many features, i.e $X=(x_1,\dots ,x_n)$
$y$	Target. Output values from the dataset. Also called: ground truth, response variables, or label
$\hat{y}$	Prediction. Output calculated/suggested by the model

Example

	Surface ($m^2$)	Number of rooms	Price / month
1	80	2	1500
2	90	2	1600
3	100	3	1750
4	110	3	1900

Dataset $S$, composed of 4 samples

• Feature $x_1$ → Surface
• Feature $x_2$ → Nbr. of rooms
• Target $y_1$ → Price / month

Architectures

Decision Tree

Decision tree is a supervised learning approach, which can be used in classification or regression tasks.