Scalable models for probabilistic forecasting with Fuzzy Time Series
Autor(a) principal: | |
---|---|
Data de Publicação: | 2019 |
Tipo de documento: | Tese |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UFMG |
Texto Completo: | http://hdl.handle.net/1843/30040 |
Resumo: | In the field of time series forecasting, the most known methods are based on pointforecasting. However, this kind of forecasting has a serious drawback: it does not quantifythe uncertainties inherent to natural and social processes neither other uncertaintiescaused by the data gathering and processing. Because this in last years the interval andprobabilistic forecasting methods have been gaining more attention of researches, speciallyon environmental and economical sciences. But these techniques also have their own issuesdue to the methods being black-boxes and requiring stochastic simulations and ensemblesof multiple forecasting methods which are computationally expensive.On the other hand, the data volume (number of instances) and dimensionality (numberof variables) have reached magnitudes even greater, due to the commoditizing of thecapturing and storing computational devices, in a phenomenon known as Big Data. Suchfactors impact directly on the model’s training and updating costs, and for time serieswith Big Data characteristics, the scalability became a decisive factor in the choosing ofpredictive methods.In this context the Fuzzy Time Series (FTS) methods emerge, which have been growing inrecent years due to their accurate results, easiness of implementation, low computationalcost and model explainability. The Fuzzy Time Series methods have been applied toforecast electric load, market assets, economical indicators, tourism demand etc. But thereis a lack on FTS literature regarding interval and probabilistic forecasting.This thesis proposes new scalable Fuzzy Time Series methods and discusses its applicationto point, interval and probabilistic forecasting of mono and multivariate time series, for oneto many steps ahead. The parameters and hyper-parameters are discussed and fine tunningalternatives are presented. Finally the proposed methods are compared with the mainFuzzy Time Series techniques and other literature approaches using environmental andstock market data. The proposed methods obtained promising results on point, intervaland probabilistic forecasting and presented low computational cost, making it useful for awide range of applications. |
id |
UFMG_896e94bda15f23002d9c3d91b04d2d24 |
---|---|
oai_identifier_str |
oai:repositorio.ufmg.br:1843/30040 |
network_acronym_str |
UFMG |
network_name_str |
Repositório Institucional da UFMG |
repository_id_str |
|
spelling |
Frederico Gadelha Guimarãeshttp://lattes.cnpq.br/2472681535872194Hossein Javedani SadaeiFrederico Gadelha GuimarãesHossein Javedani SadaeiGuilherme de Alencar BarretoGraçaliz Pereira DimuroMarcos Flávio Silveira Vasconcelos D Angelohttp://lattes.cnpq.br/2433080030239869Petrônio Cândido de Lima e Silva2019-09-17T19:20:06Z2019-09-17T19:20:06Z2019-09-02http://hdl.handle.net/1843/30040In the field of time series forecasting, the most known methods are based on pointforecasting. However, this kind of forecasting has a serious drawback: it does not quantifythe uncertainties inherent to natural and social processes neither other uncertaintiescaused by the data gathering and processing. Because this in last years the interval andprobabilistic forecasting methods have been gaining more attention of researches, speciallyon environmental and economical sciences. But these techniques also have their own issuesdue to the methods being black-boxes and requiring stochastic simulations and ensemblesof multiple forecasting methods which are computationally expensive.On the other hand, the data volume (number of instances) and dimensionality (numberof variables) have reached magnitudes even greater, due to the commoditizing of thecapturing and storing computational devices, in a phenomenon known as Big Data. Suchfactors impact directly on the model’s training and updating costs, and for time serieswith Big Data characteristics, the scalability became a decisive factor in the choosing ofpredictive methods.In this context the Fuzzy Time Series (FTS) methods emerge, which have been growing inrecent years due to their accurate results, easiness of implementation, low computationalcost and model explainability. The Fuzzy Time Series methods have been applied toforecast electric load, market assets, economical indicators, tourism demand etc. But thereis a lack on FTS literature regarding interval and probabilistic forecasting.This thesis proposes new scalable Fuzzy Time Series methods and discusses its applicationto point, interval and probabilistic forecasting of mono and multivariate time series, for oneto many steps ahead. The parameters and hyper-parameters are discussed and fine tunningalternatives are presented. Finally the proposed methods are compared with the mainFuzzy Time Series techniques and other literature approaches using environmental andstock market data. The proposed methods obtained promising results on point, intervaland probabilistic forecasting and presented low computational cost, making it useful for awide range of applications.No campo da previsão de séries temporais os métodos mais difundidos baseiam-se em predição por ponto. Esse tipo de previsão, no entanto, tem um sério inconveniente: ele não quantifica as incertezas inerentes aos processos naturais e sociais nem outras incertezas decorrentes da captura e processamento dos dados. Por isso nos últimos anos os métodos de previsão intervalar e probabilística têm ganhado a atenção dos pesquisadores, particularmente nas ciências climáticas e na econometria. Mas outro inconveniente vem do fato de grande parte dos métodos de previsão probabilística serem métodos de caixa preta e demandarem simulações estocásticas ou ensembles de métodos preditivos que são computacionalmente despendiosos. Por outro lado, o volume (número de registros) e a dimensionalidade (número de variáveis) dos dados vêm alcançando magnitudes cada vez maiores, graças ao barateamento dos dispositivos computacionais de captura e armazenamento de dados, um fenômeno conhecido como Big Data. Tais fatores impactam diretamente no custo de treinamento e atualização dos modelos e, para séries temporais com essas características, a escalabilidade tornou-se um fator decisivo na escolha dos métodos preditivos. Nesse contexto emergem os métodos de Séries Temporais Nebulosas, que vêm em crescente expansão nos últimos anos dado os seus resultados acurados, a facilidade de implementação dos métodos, o seu baixo custo computacional e a interpretabilidade de seus modelos. Os métodos de Séries Temporais Nebulosas têm sido utilizados em áreas como previsão de demanda energética, indicadores e ativos de mercado, turismo entre outras. Mas há lacunas na literatura de tais métodos referentes a escalabilidade para grandes volumes de dados e previsão probabilística e por intervalos. A presente tese propõe novos métodos escaláveis de Séries Temporais Nebulosas e investiga a aplicação desses modelos na previsão por ponto, intervalar e probabilística, para uma ou mais variáveis e para mais de um passo à frente. Os parâmetros e hiperparâmetros dos métodos são discutidos e são apresentadas alternativas de ajuste fino dos modelos. Os métodos propostos são então comparados com as principais técnicas de Séries Temporais Nebulosas e outros modelos estatísticos utilizando dados ambientais e do mercado de ações. Os modelos propostos apresentaram resultados promissores tanto nas previsões por ponto quanto nas previsões por intervalo e probabilísticas e com baixo custo computacional, tornando-os úteis para um vasta gama de aplicaçõesengUniversidade Federal de Minas GeraisPrograma de Pós-Graduação em Engenharia ElétricaUFMGBrasilENG - DEPARTAMENTO DE ENGENHARIA ELÉTRICAhttp://creativecommons.org/licenses/by-nd/3.0/pt/info:eu-repo/semantics/openAccessEngenharia elétricaAnálise de séries temporaisEscalabilidadeSéries Temporais NebulosasEscalabilidadePrevisão ProbabilísticaPrevisão por IntervaloScalable models for probabilistic forecasting with Fuzzy Time SeriesModelos escaláveis para previsão probabilística com séries temporais nebulosasinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/doctoralThesisreponame:Repositório Institucional da UFMGinstname:Universidade Federal de Minas Gerais (UFMG)instacron:UFMGORIGINALFinal_Thesis.pdfFinal_Thesis.pdfapplication/pdf7930425https://repositorio.ufmg.br/bitstream/1843/30040/1/Final_Thesis.pdf698683d322773c4282afb9814c2b5089MD51CC-LICENSElicense_rdflicense_rdfapplication/rdf+xml; charset=utf-8805https://repositorio.ufmg.br/bitstream/1843/30040/2/license_rdf00e5e6a57d5512d202d12cb48704dfd6MD52LICENSElicense.txtlicense.txttext/plain; charset=utf-82119https://repositorio.ufmg.br/bitstream/1843/30040/3/license.txt34badce4be7e31e3adb4575ae96af679MD53TEXTFinal_Thesis.pdf.txtFinal_Thesis.pdf.txtExtracted texttext/plain344287https://repositorio.ufmg.br/bitstream/1843/30040/4/Final_Thesis.pdf.txt31fd20fa136dda7bc7a63bde37887336MD541843/300402020-01-24 15:46:23.0oai:repositorio.ufmg.br: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Repositório de PublicaçõesPUBhttps://repositorio.ufmg.br/oaiopendoar:2020-01-24T18:46:23Repositório Institucional da UFMG - Universidade Federal de Minas Gerais (UFMG)false |
dc.title.pt_BR.fl_str_mv |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
dc.title.alternative.pt_BR.fl_str_mv |
Modelos escaláveis para previsão probabilística com séries temporais nebulosas |
title |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
spellingShingle |
Scalable models for probabilistic forecasting with Fuzzy Time Series Petrônio Cândido de Lima e Silva Séries Temporais Nebulosas Escalabilidade Previsão Probabilística Previsão por Intervalo Engenharia elétrica Análise de séries temporais Escalabilidade |
title_short |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
title_full |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
title_fullStr |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
title_full_unstemmed |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
title_sort |
Scalable models for probabilistic forecasting with Fuzzy Time Series |
author |
Petrônio Cândido de Lima e Silva |
author_facet |
Petrônio Cândido de Lima e Silva |
author_role |
author |
dc.contributor.advisor1.fl_str_mv |
Frederico Gadelha Guimarães |
dc.contributor.advisor1Lattes.fl_str_mv |
http://lattes.cnpq.br/2472681535872194 |
dc.contributor.advisor-co1.fl_str_mv |
Hossein Javedani Sadaei |
dc.contributor.referee1.fl_str_mv |
Frederico Gadelha Guimarães |
dc.contributor.referee2.fl_str_mv |
Hossein Javedani Sadaei |
dc.contributor.referee3.fl_str_mv |
Guilherme de Alencar Barreto |
dc.contributor.referee4.fl_str_mv |
Graçaliz Pereira Dimuro |
dc.contributor.referee5.fl_str_mv |
Marcos Flávio Silveira Vasconcelos D Angelo |
dc.contributor.authorLattes.fl_str_mv |
http://lattes.cnpq.br/2433080030239869 |
dc.contributor.author.fl_str_mv |
Petrônio Cândido de Lima e Silva |
contributor_str_mv |
Frederico Gadelha Guimarães Hossein Javedani Sadaei Frederico Gadelha Guimarães Hossein Javedani Sadaei Guilherme de Alencar Barreto Graçaliz Pereira Dimuro Marcos Flávio Silveira Vasconcelos D Angelo |
dc.subject.por.fl_str_mv |
Séries Temporais Nebulosas Escalabilidade Previsão Probabilística Previsão por Intervalo |
topic |
Séries Temporais Nebulosas Escalabilidade Previsão Probabilística Previsão por Intervalo Engenharia elétrica Análise de séries temporais Escalabilidade |
dc.subject.other.pt_BR.fl_str_mv |
Engenharia elétrica Análise de séries temporais Escalabilidade |
description |
In the field of time series forecasting, the most known methods are based on pointforecasting. However, this kind of forecasting has a serious drawback: it does not quantifythe uncertainties inherent to natural and social processes neither other uncertaintiescaused by the data gathering and processing. Because this in last years the interval andprobabilistic forecasting methods have been gaining more attention of researches, speciallyon environmental and economical sciences. But these techniques also have their own issuesdue to the methods being black-boxes and requiring stochastic simulations and ensemblesof multiple forecasting methods which are computationally expensive.On the other hand, the data volume (number of instances) and dimensionality (numberof variables) have reached magnitudes even greater, due to the commoditizing of thecapturing and storing computational devices, in a phenomenon known as Big Data. Suchfactors impact directly on the model’s training and updating costs, and for time serieswith Big Data characteristics, the scalability became a decisive factor in the choosing ofpredictive methods.In this context the Fuzzy Time Series (FTS) methods emerge, which have been growing inrecent years due to their accurate results, easiness of implementation, low computationalcost and model explainability. The Fuzzy Time Series methods have been applied toforecast electric load, market assets, economical indicators, tourism demand etc. But thereis a lack on FTS literature regarding interval and probabilistic forecasting.This thesis proposes new scalable Fuzzy Time Series methods and discusses its applicationto point, interval and probabilistic forecasting of mono and multivariate time series, for oneto many steps ahead. The parameters and hyper-parameters are discussed and fine tunningalternatives are presented. Finally the proposed methods are compared with the mainFuzzy Time Series techniques and other literature approaches using environmental andstock market data. The proposed methods obtained promising results on point, intervaland probabilistic forecasting and presented low computational cost, making it useful for awide range of applications. |
publishDate |
2019 |
dc.date.accessioned.fl_str_mv |
2019-09-17T19:20:06Z |
dc.date.available.fl_str_mv |
2019-09-17T19:20:06Z |
dc.date.issued.fl_str_mv |
2019-09-02 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/doctoralThesis |
format |
doctoralThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://hdl.handle.net/1843/30040 |
url |
http://hdl.handle.net/1843/30040 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.rights.driver.fl_str_mv |
http://creativecommons.org/licenses/by-nd/3.0/pt/ info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
http://creativecommons.org/licenses/by-nd/3.0/pt/ |
eu_rights_str_mv |
openAccess |
dc.publisher.none.fl_str_mv |
Universidade Federal de Minas Gerais |
dc.publisher.program.fl_str_mv |
Programa de Pós-Graduação em Engenharia Elétrica |
dc.publisher.initials.fl_str_mv |
UFMG |
dc.publisher.country.fl_str_mv |
Brasil |
dc.publisher.department.fl_str_mv |
ENG - DEPARTAMENTO DE ENGENHARIA ELÉTRICA |
publisher.none.fl_str_mv |
Universidade Federal de Minas Gerais |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFMG instname:Universidade Federal de Minas Gerais (UFMG) instacron:UFMG |
instname_str |
Universidade Federal de Minas Gerais (UFMG) |
instacron_str |
UFMG |
institution |
UFMG |
reponame_str |
Repositório Institucional da UFMG |
collection |
Repositório Institucional da UFMG |
bitstream.url.fl_str_mv |
https://repositorio.ufmg.br/bitstream/1843/30040/1/Final_Thesis.pdf https://repositorio.ufmg.br/bitstream/1843/30040/2/license_rdf https://repositorio.ufmg.br/bitstream/1843/30040/3/license.txt https://repositorio.ufmg.br/bitstream/1843/30040/4/Final_Thesis.pdf.txt |
bitstream.checksum.fl_str_mv |
698683d322773c4282afb9814c2b5089 00e5e6a57d5512d202d12cb48704dfd6 34badce4be7e31e3adb4575ae96af679 31fd20fa136dda7bc7a63bde37887336 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFMG - Universidade Federal de Minas Gerais (UFMG) |
repository.mail.fl_str_mv |
|
_version_ |
1793890425949388800 |