In today's networked society, getting knowledge from the data becomes in a key task for many support decisions making processes. Such kind of decisions could imply serious consequences in business and life and as more information is available about the problem, more accurate will be the decision.A largely untested hypothesis of modern society is that it is important to record data as it may contain valuable information. However, we are living in the information age and it has gone into how this quantity of data might be analyzed. The number of data sources increases, which makes harder and, in some cases, unfeasible the approach of storing every piece of data. Actually, the amount of available storage capacity could be not enough in order to record all the digital info generated in the world and this could become one the main triggers for a paradigm shift. Another aspect is that it is necessary to react instantly to meaningful changes in data and detect complex patterns over time. Data stream paradigm was born as an answer to the continuous data problem that cannot be stored and analyzed in efficient manner. Data S treams can handle bigger data sizes than memory, and can extend to challenging real-time applications not previously tackled by e.g. time series analysis, traditional, databases, machine learning and data mining techniques.The assumption is that processing examples can be processed on the fly, that is, as soon as they arrive within a high speed stream, and then, must be discarded in order to make room for subsequent data (e.g. examples in a learning scenario) .In this article we will discuss the application of data stream techniques in the S mart Grid use case. The new smart grids will require real time reaction and make distribution decision very fast in order to cover unexpected demand peaks and renewable generation peaks.

No abstract