Partitioning-Clustering Techniques Applied to the Electricity Price Time Series
Clustering is used to generate groupings of data from a large dataset, with the intention of representing the behavior of a system as accurately as possible. In this sense, clustering is applied in this work to extract useful information from the electric
- PDF / 413,111 Bytes
- 10 Pages / 430 x 660 pts Page_size
- 64 Downloads / 198 Views
rea of Computer Science. Pablo de Olavide University, Spain {fmaralv,ali}@upo.es 2 Department of Computer Science. University of Seville, Spain [email protected] 3 Department of Electrical Engineering. University of Seville, Spain [email protected]
Abstract. Clustering is used to generate groupings of data from a large dataset, with the intention of representing the behavior of a system as accurately as possible. In this sense, clustering is applied in this work to extract useful information from the electricity price time series. To be precise, two clustering techniques, K-means and Expectation Maximization, have been utilized for the analysis of the prices curve, demonstrating that the application of these techniques is effective so to split the whole year into different groups of days, according to their prices conduct. Later, this information will be used to predict the price in the short time period. The prices exhibited a remarkable resemblance among days embedded in a same season and can be split into two major kind of clusters: working days and festivities. Keywords: Clustering, electricity price forecasting, time series, dayahead energy market.
1
Introduction
Due to the Spanish electricity-market deregulation, a will of obtaining optimized bidding strategies has recently arisen in the electricity-producer companies [13]. In that way, forecasting techniques are acquiring significant importance. Thus, this research lies in extracting useful information of the prices time series by using clustering techniques. In this work two well-known clustering techniques [15], K-means and Expectation Maximization (EM), are applied to prices time series in order to find those days which show a similar behavior. These labeled days will be used to forecast the day-ahead price in future work. Several forecasting techniques have already been used in forecasting miscellaneous electricity time series recently. Indeed, A. J. Conejo et al. [2] used the wavelet transform and ARIMA models and R. C. Garc´ıa et al. [4] presented a forecasting technique based on a GARCH model for this purpose. A mixing of Artificial Neural Networks and fuzzy logic were proposed in [1], while an adaptive non-parametric regression approach is handled in [17]. A model based on H. Yin et al. (Eds.): IDEAL 2007, LNCS 4881, pp. 990–999, 2007. c Springer-Verlag Berlin Heidelberg 2007
Partitioning-Clustering Techniques
991
the Weighted Nearest Neighbors methodology is presented in [14]. With the aim of dealing with the spike prices, [6] proposed a data mining approach based on both support-vector machine and probability classifier. In [5] mixed models were proposed to obtain the appropriate length of time to use for forecasting prices. However, none of them used clustering techniques applied to prices time series as a previous stage. The novel and main contribution of this paper is to apply clustering to the electricity prices time series in order to discover behavior’s patterns, as a first step to improve forecasting techniques. Therefore, this work tackle the
Data Loading...
