Short-term electric load forecasting using data mining technique

摘要

In this paper, the short-term load forecast is conducted by utilizing SARIMA model and Holt-Winters model including load classification by use of k-NN algorithm. With embodiment of a load classification procedure, it could be possible to provide more accurate load data. After load classification using 1-year training set and 1-year test set, forecast was performed through the two models. Although the differences in the results were minor, by measuring their MAPE, Holt-Winters was shown to have better performance in short-term load forecasting.