Discovering Patterns in Electricity Price Using Clustering Techniques

Abstract

Clustering is a process of grouping similar elements gathered or occurred closely together. This paper presents two clustering techniques, K-means and Fuzzy Cmeans, for the analysis of the electricity prices time series. Both algorithms are focused on extracting useful information from the data with the aim of model the time series behaviour and find patterns to improve the price forecasting. The main objective, thus, is to find a representation that preserves the original information and describes the shape of the time series data as accurately as possible. This research demonstrates that the application of clustering techniques is effective in order to distinguish several kinds of days. To be precise, two major groups can be distinguished thanks to the clustering: the first one that includes the working days and the second one that includes weekends and festivities. Equally remarkable is the similarity shown among days belonging to a same season.