Features

The discussion about RapidMiner and Weka is an opportunity to clarify the path from benchmark data sets (UCI library) to a library of ML algorithms (in Java: Weka) to an environment (RapidMiner).

The first name was NOT "Yet Another set of Learning Algorithms" (Yala), but "Yet Another Learning Environment", i.e. a workbench which supports data miners in designing new experiments with long, nested chains of preprocessing, learning, and nice visualizations of the results. You may plug-in the learning algorithm you like, new preprocessing operators, new evaluation measures, but this is not the key issue. Hence, RapidMiner is completely independent from the particular set of learning operators, e.g., the ones of Weka. You are free to use them, or use other ones.

The system offers you the easy access, the automatic flow of all processes, and persistent storage in XML. The power of the system is not measured by the number of processes a system developer could perform, of course not by the number of code lines, but by the ease of a data analyst designing the right processes for an application. It is a rapid prototyping environment for data analysts -- that's why it is called RapidMiner.

Prof. Dr. Katharina Morik,
University Dortmund,
http://www-ai.informatik.uni-dortmund.de