Adaptive Time Series Prediction Model Based on a Smoothing P-spline
- 8 Downloads
One major task of modern short-term forecasting is to increase its speed without deteriorating the quality. This is especially relevant when developing real-time forecasting models. The hybrid forecasting model proposed in this paper is based on a recurrent P-spline and enables adaptation of parameters by evolutionary optimization algorithms. An important characteristic of the proposed model is the use of a shallow prehistory. Besides, the recurrent P-spline has a cost-effective computational scheme; therefore, the forecast speed of the model is high. Simultaneous adaptation of several parameters of the P-spline allows forecast accuracy control. This leads to the creation of various versions of forecasting methods and synthesizing hybrid mathematical models with different structures.
KeywordsTime series prediction Hybrid model Evolutionary algorithms
The reported study was funded by RFBR according to the research project № 18-07-01007.
- 9.Yang, J.M.: Power system short-term load forecasting. Ph.D. thesis. Elektrotechnik und Informationstechnik der Technischen Universität, Germany, Darmstadt (2006)Google Scholar
- 13.Vapnik, V.N.: The Nature of Statistical Learning Theory. Information Science and Statistics, 2nd edn. Springer, New York (1999)Google Scholar
- 15.Zhang, M.L., Zhou, Z.H.: A k-nearest neighbor based algorithm for multi-label classification. In: 1st IEEE International Conference on Granular Computing, pp. 718–721. IEEE, New York (2005)Google Scholar
- 16.Chernoff, K., Nielsen, M.: Weighting of the k-Nearest-Neighbors. In: 20th IEEE International Conference on Pattern Recognition (ICPR), Istanbul, Turkey, pp. 666–669. IEEE (2010)Google Scholar