A1861
Title: Outlier detection from auctions in electricity markets
Authors: Luigi Grossi - University of Padova (Italy) [presenting]
Mara Sabina Bernardi - European Commission - Joint Research Center (Italy)
Andrea Cerasa - European Commission - Joint Research Centre (Italy)
Fany Nan - Joint Research Center of the European Commission Ispra (Italy)
Abstract: The goal is to detect anomalies in the energy market by analyzing the auction dynamics that underlie the formation of energy prices. Instead of solely focusing on energy prices, the entire offer curve is comprehensively examined. The approach integrates techniques from functional data analysis, time series analysis, and robust statistics to address the complexity of the data, their temporal dependencies, and potential irregular observations. Dimensionality reduction methods tailored to the features are employed that characterize offer curves, utilizing a combination of landmark registration and functional principal component analysis. Subsequently, a robust time series model is applied capable of accommodating trends, seasonality, and the influence of external covariates, all while identifying anomalies such as spikes and level shifts. For the analysis, data from the Italian market is utilized, which is provided by Gestore dei Mercati Energetici. Through the methodology, specific hours are pinpointed on particular days that exhibit anomalous behavior. Additionally, a technique is introduced to investigate these identified cases by evaluating the impact of the offers made by each operator.
