Processing Data Streams with the RapidMiner Streams Plugin

In various applications we face a plethora of data that is often growing continuously. Such data arize in monitoring settings such as server log files, manufacturing processes, sensor networks or high volume news feeds such as twitter. Analysis of such data is different to the traditional batch setting that RapidMiner initially has been designed for. In this work we present the streams library – a simple and easy to use framework to continuously process streaming data. It comes with the Streams Plugin, integrating its streaming capabilities into the RapidMiner suite. We give an overview of the architecture of the streams library and its RapidMiner integration and demonstrate its usefulness for processing very large and continuous data in several use cases.