A methodology for detection of causal relationships between discrete time series on systems
The need for detecting causality relations of process, events or variables is present in many areas of knowledge, e.g., distributed computing, the stock market, industry and medical sector. This occurs because the knowledge of these relations can often be helpful in solving a variety of problems....
Uloženo v:
Hlavní autor: | |
---|---|
Další autoři: | |
Médium: | masterThesis |
Jazyk: | pt_BR |
Vydáno: |
Brasil
|
Témata: | |
On-line přístup: | https://repositorio.ufrn.br/jspui/handle/123456789/26828 |
Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo otaguje tento záznam!
|
id |
ri-123456789-26828 |
---|---|
record_format |
dspace |
institution |
Repositório Institucional |
collection |
RI - UFRN |
language |
pt_BR |
topic |
Detecção de relações de causalidade Transferência de entropia Algoritmo K2 Redes Bayesianas CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA |
spellingShingle |
Detecção de relações de causalidade Transferência de entropia Algoritmo K2 Redes Bayesianas CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA Abreu, Rute Souza de A methodology for detection of causal relationships between discrete time series on systems |
description |
The need for detecting causality relations of process, events or variables is present
in many areas of knowledge, e.g., distributed computing, the stock market, industry and
medical sector. This occurs because the knowledge of these relations can often be helpful
in solving a variety of problems. For example, maintaining the consistency of replicated
databases when writing distributed algorithms or optimizing the purchase and sale of
stocks in the stock market. In this context, this dissertation proposes a new methodology
for detecting causality relations in systems by using information criteria and Bayesian
networks to generate the most probable structure of connections between discrete time
series. Modeling the system as a directed graph, in which the nodes are the discrete
time series and the edges represent the relations, the main idea of this work is to detect
causality relations between the nodes. This detection is made using the method of transfer
entropy, which is a method to quantify the information transferred between two variables,
and the K2 algorithm: a heuristic method whose objective is to find the most probable
belief-network structure, given a data set. Because K2 depends on the premise of having
a previous structure that defines the hierarchy among the network nodes, it is proposed in
the methodology the creation of the previous ordering on the nodes considering direct and
indirect relations, and the modeling of these relations according to the lag between cause
and effect. In addition, knowing that the K2 algorithm considers that each case of the data
set occurs simultaneously, the proposed methodology modifies the original algorithm by
inserting the dynamics of these lags into it. This modification provides a mechanism for
comparing direct and indirect causality relations regarding its contribution to the structure.
As the result, it is obtained a graph of causality relations between the series, with the
relation’s lags being explicit. |
author2 |
Oliveira, Luiz Affonso Henderson Guedes de |
author_facet |
Oliveira, Luiz Affonso Henderson Guedes de Abreu, Rute Souza de |
format |
masterThesis |
author |
Abreu, Rute Souza de |
author_sort |
Abreu, Rute Souza de |
title |
A methodology for detection of causal relationships between discrete time series on systems |
title_short |
A methodology for detection of causal relationships between discrete time series on systems |
title_full |
A methodology for detection of causal relationships between discrete time series on systems |
title_fullStr |
A methodology for detection of causal relationships between discrete time series on systems |
title_full_unstemmed |
A methodology for detection of causal relationships between discrete time series on systems |
title_sort |
methodology for detection of causal relationships between discrete time series on systems |
publisher |
Brasil |
publishDate |
2019 |
url |
https://repositorio.ufrn.br/jspui/handle/123456789/26828 |
work_keys_str_mv |
AT abreurutesouzade amethodologyfordetectionofcausalrelationshipsbetweendiscretetimeseriesonsystems AT abreurutesouzade methodologyfordetectionofcausalrelationshipsbetweendiscretetimeseriesonsystems |
_version_ |
1773961398855925760 |
spelling |
ri-123456789-268282019-05-26T06:18:34Z A methodology for detection of causal relationships between discrete time series on systems Abreu, Rute Souza de Oliveira, Luiz Affonso Henderson Guedes de Silva, Ivanovitch Medeiros Dantas da Martins, Rodrigo Siqueira Detecção de relações de causalidade Transferência de entropia Algoritmo K2 Redes Bayesianas CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA The need for detecting causality relations of process, events or variables is present in many areas of knowledge, e.g., distributed computing, the stock market, industry and medical sector. This occurs because the knowledge of these relations can often be helpful in solving a variety of problems. For example, maintaining the consistency of replicated databases when writing distributed algorithms or optimizing the purchase and sale of stocks in the stock market. In this context, this dissertation proposes a new methodology for detecting causality relations in systems by using information criteria and Bayesian networks to generate the most probable structure of connections between discrete time series. Modeling the system as a directed graph, in which the nodes are the discrete time series and the edges represent the relations, the main idea of this work is to detect causality relations between the nodes. This detection is made using the method of transfer entropy, which is a method to quantify the information transferred between two variables, and the K2 algorithm: a heuristic method whose objective is to find the most probable belief-network structure, given a data set. Because K2 depends on the premise of having a previous structure that defines the hierarchy among the network nodes, it is proposed in the methodology the creation of the previous ordering on the nodes considering direct and indirect relations, and the modeling of these relations according to the lag between cause and effect. In addition, knowing that the K2 algorithm considers that each case of the data set occurs simultaneously, the proposed methodology modifies the original algorithm by inserting the dynamics of these lags into it. This modification provides a mechanism for comparing direct and indirect causality relations regarding its contribution to the structure. As the result, it is obtained a graph of causality relations between the series, with the relation’s lags being explicit. A necessidade de detectar relações de causalidade entre processos, eventos ou variáveis está presente em diversas áreas do conhecimento, por exemplo, computação distribuída, mercado de ações, indústria, medicina, etc. Isso ocorre porque a identificação dessas relações pode, muitas vezes, ser útil na solução de diversos problemas. Por exemplo, manter a consistência de bancos de dados replicados ao escrever algoritmos distribuídos ou otimizar a compra e venda de ações no mercado financeiro. Neste contexto, esta dissertação propõe uma nova metodologia para detecção de relações de causalidade em sistemas utilizando critérios de informação e redes Bayesianas para gerar uma estrutura de conexões entre séries temporais, de tempo discreto, mais provável. Modelando o sistema como um grafo, no qual os nós são as séries temporais discretas e as arestas representam as relações, a ideia principal deste trabalho é detectar relações de causalidade entre os nós. Essa detecção é feita usando o método de transferência de entropia, que é um método para quantificar a transferência de informação entre duas variáveis, e o algoritmo K2, um método heurístico cujo objetivo é encontrar a estrutura de rede Bayesiana mais provável, dado um conjunto de dados. Porque o K2 depende da premissa de ter uma estrutura previa que define a hierarquia entre os nós da rede, é proposto na metodologia a criação desta pré-ordem considerando as relações diretas e indiretas, e a modelagem destas de acordo com o atraso entre causa e efeito. Além disso, sabendo que o algoritmo K2 considera que cada instância do conjunto de dados ocorre simultaneamente, a metodologia proposta modifica o algoritmo original inserindo nele a dinâmica desses atrasos. Esta modificação provê um mecanismo para comparar as relações de causalidade direta e indireta em relação à contribuição destas para a estrutura da rede. Como resultado obtém-se um grafo que representa relações de causalidade entre as séries, com os atrasos das relações explicitadas. 2019-04-04T21:24:12Z 2019-04-04T21:24:12Z 2019-01-25 masterThesis ABREU, Rute Souza de. A methodology for detection of causal relationships between discrete time series on systems. 2019. 65f. Dissertação (Mestrado em Engenharia Elétrica e de Computação) - Centro de Tecnologia, Universidade Federal do Rio Grande do Norte, Natal, 2019. https://repositorio.ufrn.br/jspui/handle/123456789/26828 pt_BR Acesso Aberto application/pdf Brasil UFRN PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA ELÉTRICA E DE COMPUTAÇÃO |