"RapidMiner in a Cluster"

pedrohml

Hi, I'm Pedro,

I would like to know how is the development to support parallelized processing in a cluster. I have more than 100GB of text to process and a cluster with 32 machines available.

Regards.

land

Hi Pedro,
one quite unsolved problem on machine learning is, that all the algorithms to build models are nearly unparallelizable. And since they are at least quadratic in runtime not applicable on all your data, even with 3000 machines.
But if you want to classify this amount of text, then you should train your model on one machine. The application of this model is highly independent and can be done even without any cluster structure by simply starting the application process on a subset of the data.

Greetings,
  Sebastian