New: RapidMiner 4.3

[Ingo Mierswa]

Hello!
 
New Release: RapidMiner 4.3

We are happy to announce the release of RapidMiner 4.3. During the last months, we worked hard on improving RapidMiner to meet your data analysis, ETL, and BI requirements. The result is now available for download and provides more than 50 new operators and a lot of exciting new features:

• Improve your analysis and data transformations by upgraded data pivotings, new aggregation functions, or date and time handling
• Enjoy the simplified usage of function based attribute construction, optimized wizards, or the new visualizations including zooming and panning
• Create more powerful processes with enhanced macros or the new result storage mechanism
• Make use of the improved lift charts and optimize your marketing, your churn reduction efforts, your...

You can find detailed information about RapidMiner 4.3 at

http://rapid-i.com/content/view/133/1/


Download

The new version and all RapidMiner plugins are  available at
 
http://rapid-i.com
 

Enterprise Edition

Of course, the RapidMiner Enterprise Edition benefits from all improvements of the new version. In addition, the reporting engine was extended and provides now several new parameters for reporting as well as new parallized variants of analysis operators were also included. The new Decision Tree Learner, for example, is now able to learn the model 8 times faster than the usual version on a 8 core machine.  Check out the details at the Enterprise Edition feature page.


Please post comments and requests in this thread.
 
Cheers,
Ingo

[steffen]

Hello,

some remarks from my side:

My personal sources of joy

• Macros are able to handle numerical values
• More operators to handle nominal values
• An easy-to-use AttributeConstruction
• Showing parameters depending on the setters of other parameters
• all the fantastic stuff I have not found yet
• No comment 

Feature Requests

• Not a real feature request, see this thread: http://rapid-i.com/rapidforum/index.php/topic,280.0.html. As far as I know nominal attributes are not able to handle non-binary non-ordinal nominal values
• I was a little misguided by "enhanced macro definition". Maybe it is possible in the future  to allow the definition of arbitrary functions to set a macro value.  But I am afraid we are going to slip back into the script discussion. The number of use cases I can imagine right now are ... manageable, but such an extension would promote macros to what they really should be: "Variables" to "code" processes.

regards,

Steffen

PS: (Although noone is interested in this:) Due to other projects I have to reduce the already reduced "number of anwers per week"  to nearly 0. But I will come back from time to time to see if they are interesting discussions / projects.  Have fun and keep up the great work !

[Ingo Mierswa]

Hi Steffen,

thanks for your remarks. And since I mentioned it only once in another topic and want to give the kudos: the idea for the IOStorer and IORetriever was one Steffen mentioned to us when we met this summer. And it is a real great idea allowing for much more powerful processes. Thanks again!

About the macro definition: you are right, the enhanced refers to macro usage in numerical parameters and parameter lists as well as to the new macro definition operator "DataMacroDefinition" (very useful  ) and the new operator "Macro2Log" together with "ProcessLog2Data". They give you a lot of options for data collection, altering, more flexible processes etc.

The next step is logical but we really did not want to wait longer for the release: allowing expression parsing also for macros. And yes, from my point of views this is very close to scripts but more, well, self-contained than changing to a different language.

PS: (Although noone is interested in this:) Due to other projects I have to reduce the already reduced "number of anwers per week"  to nearly 0. But I will come back from time to time to see if they are interesting discussions / projects.  Have fun and keep up the great work !

That's a pity! 
You are certainly one of our users and subscribers to this (and the old  ) forum with a lot of valuable information for other (and much less experienced) users. And you were always a source for great feature requests and comments. I hope that you will find the time to visit us as often as possible and all the best to you.

Cheers,
Ingo