You are viewing the RapidMiner Studio documentation for version 7.6 - Check here for latest version

What's New in RapidMiner Studio 6.4?

This page describes the new features of RapidMiner Studio 6.4 as well as its enhancements and bug fixes.

A new method of workflow annotation

Collaboration among stakeholders is key for analytics initiatives and projects. With the new workflow annotation capabilities of RapidMiner Studio, you can now annotate RapidMiner processes using stickers on the Process view canvas. These stickers can be freely placed and re-sized anywhere on the canvas, including attached to individual operators.

With this tool, you can easily and visually document whole analytic processes, highlighted parts of a process, or individual steps within a process--as you build. These capabilities greatly improve collaboration among users as well as ease and streamline the maintenance and auditing of analytic processes. The new workflow annotation feature replaces the old process and operator commenting functionality. Any existing process or operator comment is automatically converted into workflow annotations when loading a process.

Some sample uses:

New extensions

RapidMiner Studio 6.4 introduces the following new and improved extensions.

Improved R integration

RapidMiner Studio 6.4 features an improved integration of the well-adopted statistical programming language R. The integration focuses on providing the core functionality needed when combining RapidMiner with R. Now, you can execute R code from within a RapidMiner process, passing data to R and passing the result of the R code execution back to RapidMiner after executing the R script. The integration has been completely revised, resulting in not only an easier installation and configuration in RapidMiner Studio and RapidMiner Server, but also in a more stable and secure integration solution. The R integration is delivered as a new extension called R Scripting, which supersedes the earlier R Extension.

Known limitation: Communication with R is limited to the system default encoding. Characters not representable in this encoding will break when processing in R.

Python integration

Analogous to the R integration, RapidMiner Studio 6.4 introduces integration with the data scientist-friendly Python programming language. You can now easily integrate Python code into your RapidMiner processes. As with R, data can be passed seamlessly from RapidMiner to Python, where it can be manipulated and used for model building or charting; Python results can then be transferred back and made available in RapidMiner.

Splunk connector

RapidMiner now provides native connectivity to Splunk, a platform for storing, searching, monitoring, and analyzing machine-generated data. With the RapidMiner Studio 6.4 connector operator, you can now build Splunk data ingestion into RapidMiner processes for deeper analysis.

Extension development kit

RapidMiner Studio 6.4 makes it much simpler to develop new extensions. First, we provide an extension template on Github that users can easily clone. Using Gradle as a modern build tool, we then provide scaffolding capabilities to quickly create a new extension stub. Also provided is documentation on how to: extend RapidMiner, implement specific operators, make use of RapidMiner's data structures, and more.

One-click extension installation

With RapidMiner Studio 6.4, you can install extensions directly from the RapidMiner Marketplace website with a single click. Each Extension page displays a button that, when clicked, starts up RapidMiner Studio and then the automatic extension installation.

New Mac version of RapidMiner Studio

The RapidMiner Studio 6.4 Mac download contains an installer app that significantly eases and accelerates Mac installation. RapidMiner Studio now feels and behaves like a native Mac application.

Enhancements and bug fixes

The following improvements are part of RapidMiner Studio 6.4.

Enhancements

Improved Process history view
Connections to RapidMiner Server no longer require equal license editions for Studio and Server. For example, professional-level RapidMiner Studio can now connect to Enterprise-level RapidMiner Server.
Improved visual feedback for port and connection interactions in the Process view
Drastically improved Process view performance
Cleaned up right-click context menu in the Process view
RapidMiner Server connections are now editable in RapidMiner Studio
Breakpoints in subprocesses are now indicated in the top right corner of the Process view
Dragging multiple repository entries into a process is now possible
Updated keyboard shortcuts and mouse handling improves Mac user experience
Ctrl + Backspace is now available for text inputs and deletes an entire word instead of a single character
On opening, problem display only occurs if a critical problem was detected
In Select Attribute operators, numeric conditions now ignore blank spaces
Improved error message shown when class weights are specified for classes that do not exist
Added display of release platform to the About screen
Unmanaged extensions are now also loaded from ~/.RapidMiner/extensions if not specified otherwise in Preferences
All sample processes have been updated and improved to be compatible with the current version
Added new sampling type of automatic to the X-Validation operator
Operator search only expands groups with hits inside
Operator search is case sensitive when search term starts with an upper case letter
API: Added draw decorator and event hooks for the Process view. See ProcessRendererView#addDrawDecorator() and ProcessRendererView#addEventDecorator().

Bug fixes

BUGFIX: Safemode dialog on startup is no longer sometimes hidden behind other windows
BUGFIX: Update Database now closes database connections after finishing
BUGFIX: Restarting after activating a license with more memory now correctly increases available memory on Windows
BUGFIX: A more meaningful error message is displayed when an invalid numeric condition is entered as a parameter
BUGFIX: Adding new database drivers via the Manage Database Drivers dialog no longer requires a restart
BUGFIX: Fixed rare error that could prevent the Manage Database Connections dialog from opening
BUGFIX: Fixed broken parameter help content for some operator parameters
BUGFIX: Calculation of a SOM-plot can now be cancelled
BUGFIX: It is no longer possible to drag operators out of the Process view
BUGFIX: Fixed rare error that could occur during automatic operator port connection
BUGFIX: Scrolling speed in the Process view is increased
BUGFIX: Fixed duplicate entry error in the History view
BUGFIX: Fixed Guess Types operator which occasionally took only the last numerical value into account
BUGFIX: A more meaningful error message is displayed when using Add generated primary keys for writing to MSSQL databases
BUGFIX: Fixed broken Execute Process operator help
BUGFIX: Disabled zoom functionality in Histogram Charts
BUGFIX: A more meaningful error message is displayed when using the Hyper Hyper operator with invalid input
BUGFIX: Principal Component Analysis operator works when applied on special attributes with missing values
BUGFIX: Fixed Read Excel operator encoding errors on Windows 8.1
BUGFIX: In Excel import wizard, wrong-typed values are parsed as missing instead of causing an error
BUGFIX: Removed unused parameter attribute type from Discretize by User Specification operator
BUGFIX: Fixed some broken templates and sample processes
BUGFIX: Clustering models now work with special attributes that contain missing values
BUGFIX: K-Medoids operator now always uses the selected measure type
BUGFIX: Fixed rare cases of broken standard coefficients for Linear Regression operator
BUGFIX: Right-clicking an operator now selects it before opening the popup menu (Linux/Mac)
BUGFIX: When installing extensions from Marketplace, dependencies are only added if not yet installed
BUGFIX: Marketplace dialogs now always open in the correct order
BUGFIX: The date functions of Generate Attributes operator now add correct metadata for new attributes
BUGFIX: Operator text parameter dialogs (e.g., the SQL query dialog) can now be closed by pressing Ctrl + Enter
BUGFIX: The log level of the Log view is now correctly restored on each start

Categories

Versions