We use one year's subset to train the Support Vector Machines (SVM) and the next year's data was used for testing with Kernel Principal Components Analysis (KPCA). This is clearly not optimal for a non-stationary time series such as we have here nevertheless the MAPE of peak load data set with back-propagation neural network [Chuang et al., 1998] is 3.0 and Support Vector Machine is 2.6. © 2010 Springer-Verlag Berlin Heidelberg.
CITATION STYLE
Liu, F. T., Chen, C. H., Chuang, S. J., & Ou, T. C. (2010). The power load forecasting by kernel PCA. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 6422 LNAI, pp. 411–424). https://doi.org/10.1007/978-3-642-16732-4_44
Mendeley helps you to discover research relevant for your work.