Feature selection and extraction along with electricity price forecasting using big data analytics

Abstract

The most important part of the smart grid (SG) is prediction of electricity price and by this prediction SG becomes cost efficient. To tackle with large amount of data in SG, it is a challenging task for existing techniques to accurately predict the electricity price. So, to handle the above mentioned problem, a framework has been proposed with three different steps: feature selection, feature extraction and classification. The purpose of feature selection is to remove irrelevant data by using extra tree classifier on the basis of pearson correlation coefficient. Feature extraction is performed using t-distributed stochastic neighbor embedding method to reduce redundancy from the selected data. For accurate electricity price forecasting, support vector machine classifier is used. Simulation results show that the proposed framework outperforms than the other methods.