Matrizes de Markov (IST)

Matrizes de Markov (IST)

Curso sobre a probabilidade de estar sol ou nuvens num determinado dia, a distribuição da população entre a cidade e os arredores ou como funciona o algoritmo de ordenação de páginas web do Google. Neste curso sobre matrizes de Markov o ênfase principal vai ser dado aos modelos de aplicação das cadeias de Markov, em que as matrizes são um dos principais protagonistas.
Nota: Esta edição do curso em modo self paced terminará no dia 28 de fevereiro de 2022.

Os outros protagonistas são os vetores de estado, que contêm, por exemplo a informação sobre a probabilidade de estar sol ou nuvens num determinado dia, ou ainda a distribuição da população entre a cidade e os arredores numa dada região.

Um dos objetivos deste curso será descrever e simular o procedimento do Google, PageRank, enquanto motor de busca, para ordenar por importância as páginas da Internet quando se faz uma pesquisa sobre um determinado tópico. Com este fim, vamos usar conceitos e propriedades de matrizes e vetores, que são objetos matemáticos associados a Álgebra Linear, mas sobre os quais não precisamos de ter muitos conhecimentos a priori.
No final deste curso, os(as) participantes estarão aptos(as) a:

  • construir modelos simples que permitem fazer certos tipos de previsões de tempo e de ocupações de salas num labirinto, entre outros modelos;
  • explicar como funciona o algoritmo de ordenação de páginas web do Google;
  • saber verificar em que condições (matemáticas) é que uma cadeia de Markov converge para um vetor de estado único (vetor estacionário).

Será de esperar ainda que no final do curso, os participantes saibam usar um software de cálculo numérico para fazer as contas mais trabalhosas destes modelos. Todos os exemplos neste curso correm no software Mathematica que é uma ferramenta de cálculo algébrico.

Pré-requisitos:

  • Ao nível de conhecimentos na área de matemática:

. pressupõe-se que o(a) participante tem alguma familiaridade com cálculos algébricos simples;
. facilita ter alguns conhecimentos de operações algébricas com matrizes e vetores.

  • Ao nível de utilização de software:

. Mathematica para algumas operações matriciais, mas podem ser usados outros programas (Maple, MATLAB, etc.), calculadoras programáveis com operações matriciais ou recorrer ao motor de busca computacional Wolfram Alpha (Wolfram|Alpha);
. as simulações deste curso correm num software gratuito (Wolfram CDF Player), que pode ser descarregado do site da Wolfram.
Os conceitos a abordar são:

  • cadeias de Markov;
  • matrizes de transição, i.e. matrizes de Markov;
  • vetores estacionários integrados em modelos lineares, que permitem prever estados futuros (ou tempos de ocupação média) da cadeia, ou seja, prever o comportamento estatístico a longo prazo da cadeia.

Os 4 tópicos semanais são dedicados a:

  • introduzir os conceitos básicos necessários para prever o longo prazo;
  • descrever diferentes modelos de passeios aleatórios em labirintos e grafos (redes), que ajudam a entender o funcionamento geral das cadeias de Markov;
  • explicitar os traços gerais de funcionamento do algoritmo do PageRank. Este procedimento será fundamentado por um teorema importante (Teorema de Perron-Frobenius), que é a base matemática que explica o comportamento estatístico dos modelos analisados durante o curso.

——————————————————————————————————————————————————
In this course on Markov matrices the main emphasis will be given to models for the application of Markov chains , where arrays are one of the main protagonists. The other players are the state vectors that contain, for example information about the probability of being sun or clouds on a given day, or the distribution of population between the city and the surrounding area in a given region.
One of the objectives of this course will describe and simulate the Google procedure, PageRank , while search engine to sort by important web pages when you do a search on a particular topic. To this end, we will use concepts and properties of matrices and vectors, which are mathematical objects associated with linear algebra, but on which we do not have much knowledge a priori .
At the end of this course, you will be able to:

  • Build simple models that allow to predict the weather and occupation times for rooms in a maze, among other models;
  • Explain how the Google search engine algorithm works in order to rank the importance of web pages;
  • Know how to verify under which (mathematical) conditions a Markov chain converges to a unique vector state (steady-state vector).

It will be expected that at the end of the course you will be capable of using a numerical software to do the more complicated calculations of these models.
Prerequisites
In what concerns mathematical background:

  • We assume that you are familiar with basic algebraic calculations;
  • The knowledge of algebraic operations on matrices and vectors helps.

In what concerns numerical software:

  • Mathematica software to perform several matrix operations, but it is possible to use other softwares (Maple, MATLAB, etc.), programmable calculators with matrix capabilities or to use the computational search engine Wolfram Alpha (Wolfram|Alpha);
  • Simulations of this course run on a free software Wolfram CDF Player) that can be downloaded from Wolfram website.

The concepts to address are:

  • Markov chains;
  • Transition matrices, i.e. Markov matrices;
  • Steady-state vectors integrated in linear models, which allow to predict future states (or occupation times) of the chain, that is, to predict the statistical behavior on the long-run of the chain.

The 4 weekly topics are dedicated to:

  • Introduce the basic concepts needed to predict the long-run;
  • Describe for a while different models of random walks in mazes and graphs (webs), which help to understand the general functioning of Markov chains;
  • Explain the main features of the PageRank algorithm. This procedure will be supported by an important theorem (Perron-Frobenius theorem), which constitutes the mathematical basis for the statistical behavior of the several models analyzed during the course
Go to Class
MOOC List is learner-supported. When you buy through links on our site, we may earn an affiliate commission.

Related Courses

Deep Learning and Reinforcement Learning (Coursera) Coursera
IBM

Deep Learning and Reinforcement Learning (Coursera)

This course introduces you to two of the most sought-after disciplines in Machine Learning: Deep Learning and Reinforcement Learning. Deep Learning is a subset of Machine Learning that has applications in both Supervised and Unsupervised Learning, and is frequently used to power most of the AI applications that we use on a daily basis. First you will learn about the theory behind Neural Networks, which are the basis of Deep Learning, as well as several modern architectures of Deep Learning.

Oct 12th 2026
5-12 Weeks
Mathematics for Machine Learning: Linear Algebra (Coursera) Coursera
Imperial College London

Mathematics for Machine Learning: Linear Algebra (Coursera)

In this course on Linear Algebra we look at what linear algebra is and how it relates to vectors and matrices. Then we look through what vectors and matrices are and how to work with them, including the knotty problem of eigenvalues and eigenvectors, and how to use these to solve problems. Finally we look at how to use these to do fun things with datasets - like how to rotate images of faces and how to extract eigenvectors to look at how the Pagerank algorithm works.

Oct 5th 2026
5-12 Weeks
Big Data Fundamentals (edX) EdX
University of Adelaide,AdelaideX

Big Data Fundamentals (edX)

Learn how big data is driving organisational change and essential analytical tools and techniques, including data mining and PageRank algorithms. Organizations now have access to massive amounts of data and it’s influencing the way they operate. They are realizing in order to be successful they must leverage their data to make effective business decisions.

Self Paced
Self-Paced
Advanced Linear Algebra: Foundations to Frontiers (edX) EdX
University of Texas at Austin,UTAustinX

Advanced Linear Algebra: Foundations to Frontiers (edX)

Learn advanced linear algebra for computing. Linear algebra is one of the fundamental tools for computational and data scientists. In Advanced Linear Algebra: Foundations to Frontiers (ALAFF), you will build your knowledge, understanding, and skills in linear algebra, practical algorithms for matrix computations, and the analysis of the effects of floating-point arithmetic as performed by computers.

Self Paced
Self-Paced
Come Rain or Shine: Understanding the Weather (FutureLearn) FutureLearn
University of Reading

Come Rain or Shine: Understanding the Weather (FutureLearn)

Understand and explore the physical processes behind the weather. Weather affects our lives almost every day through what we wear, what we eat and what we do. But why is it rainy, windy or sometimes even sunny? Explore some of the physical processes driving UK weather systems and get hands on in the world of weather with practical activities and fieldwork.

Self Paced
3 Weeks
Digital Skills: Digital Marketing (FutureLearn) FutureLearn
Accenture

Digital Skills: Digital Marketing (FutureLearn)

Learn about some of the different digital marketing strategies and techniques that are available for businesses to use. Digital marketing can be a crucial tool to grow a business. On this course you will explore what digital marketing is, why it is important and look at some digital marketing strategies including display advertising, pay per click advertising, search engine optimisation and email marketing.

Self Paced
2 Weeks
Valores Próprios (IST) MOOC Técnico
Instituto Superior Técnico, Universidade de Lisboa

Valores Próprios (IST)

Curso sobre valores próprios e vetores próprios que assenta na análise deste conceito de Álgebra Linear no contexto de transformações lineares, sistemas dinâmicos e outros modelos. No curso serão ainda bordados valores singulares. Neste curso sobre valores próprios (va.p) e vetores próprios (ve.p.) damos especial atenção às várias aplicações deste conceito fundamental de Álgebra Linear na ciência e engenharia.

Self Paced
Self-Paced
Introduction to Google SEO (Coursera) Coursera
University of California, Davis

Introduction to Google SEO (Coursera)

Ever wonder how major search engines such as Google, Bing and Yahoo rank your website within their searches? Or how content such as videos or local listings are shown and ranked based on what the search engine considers most relevant to users? Welcome to the world of Search Engine Optimization (SEO). This course is the first within the SEO Specialization and it is intended to give you a taste of SEO with some fun practices to get seen in Google.

Oct 5th 2026
4 Weeks
Generalized Linear Models and Nonparametric Regression (Coursera) Coursera
University of Colorado Boulder

Generalized Linear Models and Nonparametric Regression (Coursera)

In the final course of the statistical modeling for data science program, learners will study a broad set of more advanced statistical modeling tools. Such tools will include generalized linear models (GLMs), which will provide an introduction to classification (through logistic regression); nonparametric modeling, including kernel estimators, smoothing splines; and semi-parametric generalized additive models (GAMs). Emphasis will be placed on a firm conceptual understanding of these tools.

Oct 12th 2026
4 Weeks
Build Regression, Classification, and Clustering Models (Coursera) Coursera
CertNexus

Build Regression, Classification, and Clustering Models (Coursera)

In most cases, the ultimate goal of a machine learning project is to produce a model. Models make decisions, predictions—anything that can help the business understand itself, its customers, and its environment better than a human could. Models are constructed using algorithms, and in the world of machine learning, there are many different algorithms to choose from. You need to know how to select the best algorithm for a given job, and how to use that algorithm to produce a working model that provides value to the business. This third course within the Certified Artificial Intelligence Practitioner (CAIP) professional certificate introduces you to some of the major machine learning algorithms that are used to solve the two most common supervised problems: regression and classification, and one of the most common unsupervised problems: clustering.

Oct 5th 2026
5-12 Weeks
Advanced Linear Models for Data Science 1: Least Squares (Coursera) Coursera
Johns Hopkins University

Advanced Linear Models for Data Science 1: Least Squares (Coursera)

Welcome to the Advanced Linear Models for Data Science Class 1: Least Squares. This class is an introduction to least squares from a linear algebraic and mathematical perspective. Before beginning the class make sure that you have the following: a basic understanding of linear algebra and multivariate calculus; a basic understanding of statistics and regression models; at least a little familiarity with proof based mathematics; basic knowledge of the R programming language.

Oct 19th 2026
5-12 Weeks