Artboard 7 copy 20.jpg
Supervised and unsupervised classification: a mathematical-practical approach.

This bootcamp is focused on those individuals who are interested in taking their first steps in solving classification problems and does not require prior knowledge on these topics. It is necessary to have notions of programming or a taste for wanting to learn to program in a language.

 

The material is divided into three parts:

  1. Approach to real problems and their difficulties.

  2. Mathematical elements behind the main algorithms.

  3. Tips for implementation.

 

The practical problems that we will study are the following:

  1. Movie recommendations on Netflix.

  2. Facial recognition: children vs adults

  3. Retail analysis, how do stores choose which clothes to sell?

  4. Customer Relationship Management: segment your audience to know what to sell them and how.

Syllabus

  • From the implementation point of view, the student will learn to use the scientific method to analyze company problems and to distinguish for which hypotheses classification algorithms should be applied.

  • From a mathematical point of view we will study the following algorithms:

 

Supervised classification

  1. Linear perceptron

  2. Decision trees

 

Unsupervised classification

  1. K-nearest neighbors

  2. K-means clustering

  3. K-modes clustering

The course includes various modules focused on studying both the language and the mathematical concepts necessary to understand algorithms in depth:

Norms and forms in Euclidean geometry.

  1. Random variables

  2. Lipschitz functions

  3. Optimization

  4. Gini impurity

  5. Information gain

The entire course will be given from a practical perspective by first analyzing the business question.