Last modified: 2018-05-18
Abstract
In this paper, we describe a tool for detecting outliers in electricity prices. The tool consists primarily of a filtering procedure (Whitaker smoother) that removes unpredictable effects and captures long-term smooth variations. In order to identify outliers, we compare the observed prices with smoothed ones. If the differencebetween the two exceeds a predetermined limit, the corresponding prices are considered anomalous and candidates for an appropriate statistical treatment. The new tool is compared with another method based on local regression.
References
1. Craven, P., Wahba, G.: Smoothing noisy data with spline functions estimating the correct degree of smoothing by the method of generalized cross-validation. Numer. Math. 31, 377– 403 (1979)
2. Garcia, D.: Robust smoothing of gridded data in one and higher dimensions with missing values. Comput. Stat. Data Anal. 54, 1167–1178 (2010)
3. Hyndman, R.J. and Khandakar, Y . Automatic time series forecasting: the forecast package for R. J. Stat. Softw. 26,1–22 (2008)
4. Weron, R., Zator, M.: A note on using the HodrickPrescott filter in electricity markets. Energy Economics. 48 1–6 (2015)
5. Whittaker, E.T.: On a new method of graduation. Proc. R. Soc. Edinburgh. 44, 77–83 (1923)