A data mining paradigm to forecast weather sensitive short-term energy consumption

This paper presents an approach to forecast three-day ahead hourly electric energy consumption. This approach makes use of several significant features to come up with a precise prediction ranging from hourly electric energy usage to weather data in a predefined period of time. Once data cleaning and preprocessing are done, patterns of electric energy consumption are extracted. To extract energy consumption patterns, Neural Network and Support Vector Machine are adapted in a novel manner. Results show that the presented model achieves higher accuracy compared to the exiting approaches in power industry.