New release: on Tuesday, May 29th, 2007, Rapid-I released a new version of the open-source data mining software RapidMiner 4.0beta (formerly: YALE).

Because of legal issues, Rapid-I decided to change the name of YALE. The name RapidMiner was chosen for its fit to the company name Rapid-I and to the naming scheme of the line of products planned by Rapid-I for the future. Only the name changes. Everything else stays the same. RapidMiner / YALE will remain open-source software under GNU GPL and available to end-users free of charge and is also available under a commercial OEM license.

Changes from YALE 3.4 to RapidMiner 4.0beta

Preview

General Improvements:

• Improved overall speed: Most YALE runs now use less than 60% of the runtime needed before.
• Large API changes now better support the embedding of YALE into your own applications.
• All YALE file formats are now based on XML.
• Improved printing.
• Several bugfixes.

New Operators:

• More than 80 new operators in total, including:
  • Several operators for outlier detection
  • FPGrowth (fast and memory efficient association rule mining)
  • A huge amount of new learning and meta learning schemes
  • CostBasedThresholdLearner (also allowing to classify examples as unknown)
  • (Weighted) Bootstrapping and BootstrappingValidation
  • Learning missing values (instead of simply replacing them)
  • Many new preprocessing operators like Merge, Cartesian Products, group by, aggregation, sorting, etc.
  • Writing data sets into databases now possible
  • Generic attribute subset preprocessing
  • Generic visualization of models via dimensionality reduction
  • New ANOVA matrix
• The clustering plugin is now part of the YALE core and hence does no longer need to be installed separately.

New Look and Feel:

• Drag & Drop for operator trees
• Completely revised look and feel as well as icons
• New file chooser providing favorites
• All tables (viewers) can be sorted by all columns by clicking on the corresponding table headers
• All textual results now support text selection allowing for copy and paste into other applications
• Log scale added as new option to usual scatter plotter
• Several chart plots added (new bars 2D and 3D, pie charts 2D and 3D, bubble plotter)
• Graphical User Interface (GUI) is now able to immediately stop a running experiment
• Graph view for Bayesian net models added
• Textual and graphical view modes added for models which are capable of both, e.g. decision trees and Bayesian nets
• Result history viewer showing textual descriptions of all experiment results in the session so far; allows also the calculations of Anova for different results

New Functions:

• Improved example filter (now also supporting inversion and concatenations)
• New additional performance criteria: Spearman's rho and Kendall's tau
• New data representation types based on short or even boolean further reducing the amount of memory needed
• New HSQLDB JDBC database drivers
• ExampleSetWriter now also supports zipped data files
• Source definition added for all IO objects, i.e. the results can now show, which operator was their creator, if necessary
• Improved automatic value type guessing in ExampleSource configuration wizard
• Weighted performance measures added for weighted means of the per-class recalls and precisions
• Model writing and loading now also works for zipped files (.gz)
• Improved attribute statistics handling and display
• Updatable models now supported
• XRFF format now supported

Links