| Author: | Mark Hall <mhall{[at]}pentaho.com> and Eibe Frank <eibe{[at]}cs.waikato.ac.nz> | |
| Category: | R integration | |
| Changes: | Updated library .jar files (latest xstream, jna, mxparser, and JRI/rJava 1.0-6 .jar files) and .zip files for preload libraries. MLR classifier now installs XML package when it is missing. Fixed problem in MLRClassifier when learning scheme tried to write to file at prediction time. Improved instructions for installation and update. | |
| Date: | 2022-01-12 | |
| Depends: | weka (>=3.8.1|<3.9.0) | |
| Description: | This plugin enables R functionality to be used through WEKA. It provides an R console, a Knowledge Flow component for executing an R script, and a wrapper classifier for the MLR (machine learning in R) R package. The interactive R console enables visualization of data loaded into WEKA using R. This package requires the R software to be installed on the system before it can be used. The bitness of the OS, R, and Java must match up. We recommend using 64-bit versions of all three.The plugin will attempt to set all required environment variables for the R integration. On macOS and Linux, this process assumes that the home of R on the system is /Library/Frameworks/R.framework/Resources and /usr/lib/R respectively. On Windows, R's home will be inferred automatically through the Windows registry. The plugin will attempt to download and install all required R libraries such as rJava and JavaGD automatically into R's user library folder. The location of the R user library folder is established by interrogating R directly. If R is installed in a non-standard location, the plugin can be informed about this by setting the environment variable R_HOME or the Java property r.home to point to the installation directory for R. If a non-standard destination for the R user library folder is desired, then this can be set using the R_LIBS_USER environment variable or the Java property r.libs.user. If the native libjri in R's rJava package is not found automatically, the jri.native.library java property can be set to point directly to the jri native library file, or the java.library.path can be set to point to the directory that contains the jri native library file. More information on configuring Java to access JRI under various OS can be found at https://www.rforge.net/JRI/ Example Knowledge Flow template files demonstrating the R components can be found in ${WEKA_HOME}/packages/RPlugin/templates. | |
| License: | LGPL 3.0 | |
| Maintainer: | Mark Hall <mhall{[at]}pentaho.com> and Eibe Frank <eibe{[at]}cs.waikato.ac.nz> | |
| MessageToDisplayOnInstallation: | This plugin requires the R software to be installed on the system before it can be used. The bitness of the OS, R, and Java must match up. We recommend using 64-bit versions of all three. The plugin will attempt to set all required environment variables for the R integration. On macOS and Linux, this process assumes that the home of R on the system is /Library/Frameworks/R.framework/Resources and /usr/lib/R respectively. On Windows, R's home will be inferred automatically through the Windows registry. The plugin will attempt to download and install all required R libraries such as rJava and JavaGD automatically into R's user library folder. The location of the R user library folder is established by interrogating R directly. On Linux, where R packages are compiled from source, a Java JDK with javac must be used to run the WEKA package manager to install rJava and JavaGD. Installation with the Java JRE that comes with WEKA will fail. Also, for MLRClassifier, the R XML package, along with the R mlr package, will be attempted to be installed. On Linux, for this to succeed the xml2dev library needs to be available on the system (e.g., to install this on Ubuntu, use "sudo apt install libxml2-dev"). If you have an existing R user library folder, we recommend to update all R packages using the command "update.packages(ask = FALSE)" after starting R manually. Alternatively, if none of the installed R packages is used outside WEKA, it is possible to simply remove the R user library folder manually before starting the installation of the RPlugin so that the newest library versions are installed. If R is installed in a non-standard location, the plugin can be informed about this by setting the environment variable R_HOME or the Java property r.home to point to the installation directory for R. If a non-standard destination for the R user library folder is desired, then this can be set using the R_LIBS_USER environment variable or the Java property r.libs.user. If the native libjri in R's rJava package is not found automatically, the jri.native.library java property can be set to point directly to the jri native library file, or the java.library.path can be set to point to the directory that contains the jri native library file. More information on configuring Java to access JRI under various OS can be found at https://www.rforge.net/JRI/ Example Knowledge Flow template files demonstrating the R components can be found in ${WEKA_HOME}/packages/RPlugin/templates. | |
| PackageURL: | https://prdownloads.sourceforge.net/weka/RPlugin1.2.32.zip?download | |
| URL: | https://markahall.blogspot.co.nz/2012/07/r-integration-in-weka.html | |
| Version: | 1.2.32 |