Radoop training capabilities

calippo

I am waiting for a radoop license and I would like to fully understand its capabilities wrt spark integration.
I know that it allows to train spark models train predictive models in parallel...
What about other rapidminer algorithms? is possible to map a general classifier on a spark cluster and reduce the result using baggin/dagging or whatever?

MartinLiebig

Hi,

it is not yet possible to run a usual Rapidminer learner on different nodes. For a lot of algorithms it is simply not possible to be trained in a distributed fashion. If you have a radoop licence you get new additional operators for the spark algorithms which can be used.

However - what is possible is ofc to apply a rapidminer model on hadoop. So you can learn on a sampled data set in memory and apply it in hadoop.

Cheers,
Martin