A carregar...
Publicação
Tendências dos mercados day-ahead e intraday de eletricidade no Mibel: análise de dados multivariada
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.26 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
A atual penetração de energias renováveis no portefólio de geração da commodity
eletricidade tem vindo a crescer de forma galopante ano após ano. Estes contributos
acarretam incertezas devido aos fenómenos externos como previsões meteorológicas,
pouco previsíveis e não controláveis. Por outro lado, a geração de energia elétrica está
também dependente do preço de outras commodities, como o gás, petróleo e carvão.
Como resultado tem-se assistido a uma maior volatilidade do preço da eletricidade.
O presente trabalho final de mestrado foca-se em dar respostas no que toca à
previsão do mercado à vista (spot): day-ahead ou intraday, mais vantajoso na
aquisição/venda da commodity eletricidade, a fim de maximizar as vantagens
competitivas. O mercado ibérico de eletricidade (MIBEL), mais especificamente o
espanhol foi o mercado alvo do estudo. Optou-se por usar bases de dados de acesso
público nomeadamente o esios, o mibgas e o entso-e, tendo-se realizado transformações
nas variáveis independentes e criado a da variável dependente “legenda”, que constitui
uma variável binária indicadora do mercado, day-ahead ou intraday, em que a energia
elétrica tem o preço mais reduzido. No estudo são utilizadas abordagens multivariadas de
dados tais como a Análise Discriminante, a Regressão Logística e Redes Neuronais
Artificiais. De acordo com os resultados obtidos, esta última tipologia de modelos foi o
que apresentou um valor mais elevado da taxa de acertos, em aproximadamente 70%.
Com a metodologia aqui aplicada é possível fornecer contributos para os traders
que atuam no mercado power, suportando a tomada de decisão sobre o mercado mais
vantajoso para aquisição ou venda de eletricidade no contexto atual.
The current penetration of renewables in the generation portfolio of the commodity electricity has been growing at a galloping rate year after year. These contributions entail uncertainties due to external phenomena such as meteorological forecasts, that are poorly predictable and uncontrollable. On the other hand, electrical energy generation also depends on the price of other commodities, such as gas, oil, and coal. As a result, there has been a higher volatility in the price of electricity. The present master's final work focuses on providing answers regarding the forecast of the spot market: day-ahead or intraday, more advantageous in the acquisition/sale of the electricity commodity, in order to maximize competitive advantages. The Iberian electricity market (MIBEL), more specifically the Spanish one was the target of the study. It was opted for the use of publicly accessible databases, namely esios, mibgas, and entso-e, where transformations were made in the independent variables and it was created the dependent variable "legenda", which constitutes a market indicator binary variable, day-ahead or intraday, in which electricity has the lowest price. In the study, multivariate data approaches are used, such as Discriminant Analysis, Logistic Regression, and Artificial Neural Networks. According to the results obtained, this last typology of models was the one that presented the highest value of the success rate, at approximately 70%. According to the methodology here applied it is possible to provide contributions to traders, who operate in the power market, supporting decision-making about the most advantageous market for the purchase or sale of electricity in the current context.
The current penetration of renewables in the generation portfolio of the commodity electricity has been growing at a galloping rate year after year. These contributions entail uncertainties due to external phenomena such as meteorological forecasts, that are poorly predictable and uncontrollable. On the other hand, electrical energy generation also depends on the price of other commodities, such as gas, oil, and coal. As a result, there has been a higher volatility in the price of electricity. The present master's final work focuses on providing answers regarding the forecast of the spot market: day-ahead or intraday, more advantageous in the acquisition/sale of the electricity commodity, in order to maximize competitive advantages. The Iberian electricity market (MIBEL), more specifically the Spanish one was the target of the study. It was opted for the use of publicly accessible databases, namely esios, mibgas, and entso-e, where transformations were made in the independent variables and it was created the dependent variable "legenda", which constitutes a market indicator binary variable, day-ahead or intraday, in which electricity has the lowest price. In the study, multivariate data approaches are used, such as Discriminant Analysis, Logistic Regression, and Artificial Neural Networks. According to the results obtained, this last typology of models was the one that presented the highest value of the success rate, at approximately 70%. According to the methodology here applied it is possible to provide contributions to traders, who operate in the power market, supporting decision-making about the most advantageous market for the purchase or sale of electricity in the current context.
Descrição
Palavras-chave
Análise Multivariada Mercado Eletricidade Análise Discriminante Regressão Logística Redes Neuronais Artificiais Multivariate Analysis Power Market Discriminant Analysis Logistic Regression Artificial Neural Networks
Contexto Educativo
Citação
Santos, Rodrigo José Prata dos (2024). “Tendências dos mercados day-ahead e intraday de eletricidade no Mibel: análise de dados multivariada”. Dissertação de Mestrado. Universidade de Lisboa. Instituto Superior de Economia e Gestão
Editora
Instituto Superior de Economia e Gestão