New probabilistic price forecasting models: Application to the Iberian electricity market

This article presents original Probabilistic Price Forecasting Models, for day-ahead hourly price forecasts in electricity markets, based on a Nadaraya–Watson Kernel Density Estimator approach. A Gaussian Kernel Density Estimator function is used for each input variable, which allows to calculate th...

Descripción completa

Detalles Bibliográficos
Autores: Monteiro, C. [0000-0003-4858-647X], Ramirez-Rosado, I.J. [0000-0002-5502-4232], Fernandez-Jimenez, L.A. [0000-0002-5633-4849], Ribeiro, M.
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2018
País:España
Institución:Universidad de La Rioja (UR)
Repositorio:RIUR. Repositorio Institucional de la Universidad de La Rioja
OAI Identifier:oai:portal.dialnet.es:doc/5bbc6909b750603269e813e4
Acceso en línea:https://investigacion.unirioja.es/documentos/5bbc6909b750603269e813e4
Access Level:acceso abierto
Palabra clave:Electricity prices
Iberian electricity market
Market prices
Probabilistic forecast
Short-term forecasting
Descripción
Sumario:This article presents original Probabilistic Price Forecasting Models, for day-ahead hourly price forecasts in electricity markets, based on a Nadaraya–Watson Kernel Density Estimator approach. A Gaussian Kernel Density Estimator function is used for each input variable, which allows to calculate the parameters of the probability density function (PDF) of a Beta distribution for the hourly price variable. Thus, valuable information is obtained from PDFs such as point forecasts, variance values, quantiles, probabilities of prices, and time series representations of forecast uncertainty. A Reliability Indicator is also introduced to give a measure of “reliability” of forecasts. The Probabilistic Price Forecasting Models were satisfactorily applied to the real-world case study of the Iberian Electricity Market. Input variables of these models include recent prices, power demands and power generations in the previous day, power demands in the previous week, forecasts of demand, wind power generation and weather for the day-ahead, and chronological data. The best model, corresponding to the best combination of input variables that achieves the lowest MAE, obtains one of the highest Reliability Indicator values. A systematic analysis of MAE values of the Probabilistic Price Forecasting Models for different combinations of input variables showed that as more types of input variables were considered in these models, MAE values improved and Reliability Indicator values usually increased. © 2018 Elsevier Ltd