SUNSEED project is partially funded by EC FP7 programme under grant agreement #619437. Big Data Stream Mining • Maintain summaries of the streams, sufficient to answer the expected queries about the data: • Summaries can be in various forms: clusters (flat or hierarchic, statistical aggregates, …) • Maintain a sliding window of the most recently arrived data operations on a sliding window mimic more traditional database/mining operations • Sampling ‐ obtain representative data sample (i.e., enabling to perform correctly required operations on data) • Smart sampling (x % from stream of multiple data sources; alternative ‐ take y % of selected data sources) • Similarity comparison – smart indexing • Incremental updating of predicting models M. Skrjanc, B. Kazic {Maja.Skrjanc, Blaz.Kazic}@ijs.si , Jozef Stefan Institute, Jamova ul. 39, Ljubljana, Slovenia Forecasting in Smart Grids • Types of forecasting problems: • Electricity load (short term, medium term, long term) • Renewable sources generation • Electricity prices • Costumer segmentation • Input sources: • Historical load variables: used for learning models and detecting short term trends • Meteorological data: known to be correlated with load (depends on location) • Static data: such as special calendar data (holidays, summer season), and topology of electrical grid • Methods used: • Naive approach: Localized averages, previous values. Computationally non demanding, fast, robust and easy to maintain. Can work surprisingly well. • Classical approaches: Autoregressive (ARMIA), regression‐based statistics methods. Based on historical data. Can take advantage of seasonality trends, but usually don’t include other data sources. • Computational intelligence approaches: Artificial neural networks, support vector machines. Data driven approach that can take advantage of various heterogeneous data sources. • Hybrid methods: combine two or more different approaches in order to take advantage of specific methods benefits and overcome their drawbacks.