Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

"Q: RapidMiner Ingres Bundle 5 Installation"

Few questions:
Before installing I already had installed Java SDK, Eclipse and inside it Rapidminer with few plugins. System is 32bit Windows 7.

1. I installed RapidMiner Ingres Bundle 5.0. It does not let me update plugins installed in Eclipse. It says these are already installed. Why?
2. Installed version is 5.003. If I try to update it downloads 47 MB of data and restarts to the same version.
3. How do I get 8x performance improvement from here? What do I have to do? What tutorial to cover? Please something for newbies. I would like to see mentioned performance improvement

Find more posts tagged with

AI Studio

Installation

Accepted answers

All comments

land

Hi,
do you perfom in database data mining? Otherwise you won't gain anything. Beside this, just because I'm curious: Where is it written that the bundle gives you 8x performance?

Greetings,
Sebastian

Marin

Dear Sebastian,

from what Olaf showed us at RCOMM2010 I approximated performance improvement at about 8 times, but in practice it goes to infinite considering my 2 GB of RAM.

I do not use "in-database" data mining as RapidMiner repository is faster than that. But you can see that INGRES showed http://www.openexpo.ch/fileadmin/documents/2010Bern/Slides/25_OlafLaber.pdf it can be faster than not-in-database operation and that is what I am looking for.

I misunderstood it and there are no improvements with windowing, clustering and regression in Rapidminer working with 50.000 examples and 50 features if I switch my data from RM repository to INGRES?

Anyways, any hints regarding my questions?

land

Hi,
well, any hints regarding MY questions?

Otherwise I can't help you...

Olaf's presentation was about Ingres Vectorwise which just recently was published. It is not part of the bundle, so you have to install it separately.
But neither Vectorwise nor any other database will give you an instant speedup if you just use it for loading the data into your memory and process it there as it is done if you load the data from a RapidMiner Repository.
We are currently working on real In-Databasemining using SQL statements to derive the statistical properties from a dataset an it shows, that Vector Wise outperforms other databases by a magnitude even on a relatively small dataset. We will have further testing with larger datasets, these results will be presented on the OSBI 2010.

Greetings,
Sebastian