Concept drift detection and handling

hisham_ogbah

Hello Dear,
I hop you are doing well ...

 

I need to give me some information if there is any tools in Rapidminer for detecting and handling concept drift in the underlying data distribution.

I wonder if anyone can help me in this problem.

 

I will be grateful to you.

Telcontar120

Concept drift is a fairly broad idea, so I am not entirely certain what specific application you are interested in.

 

There is a nice time series plugin for RapidMiner that you can get for free from the marketplace, you might want to check it out.  It has a number of different windowing operators that can be used as well as calculations of moving averages that might be helpful in detecting trends.

 

A quick web search also turned up this technical paper from the University of Dortmund, you might want to contact the authors to see whether they have updated their tool (referenced in this article) to work with a newer RapidMiner version, or whether they have any other suggestions.

http://www-ai.cs.uni-dortmund.de/PublicPublicationFiles/bockermann_blom_2012a.pdf

 

Regards,