Supervised vs Unsupervised learning

Supervised learning

  • It requires labeled data for training.

  • Tasks are mainly focused on getting a solution or prediction expressed as a:

    • Classification

    • Regression

Unsupervised learning

  • It does not require labeled data due to it does not perform training.

  • Tasks are mainly focused on getting insights or guesses through:

    • Clustering

    • Visualization

    • Dimensionality reduction

    • Asociation mining

An example: KNN vs kMeans

K-Nearest Neighbors

  • Supervised algorithm

  • Based on how similar is an instance from its k neighbors

  • There is no training

  • It measures the distance between instances (Euclidian, Manhattan...)

kNN

kMeans

  • Unsupervised algorithm

  • Clustering algorithm

  • Iteratively assigns data to k groups

kMeans

Last updated