weka
| Description | ||
|
Weka is a comprehensive toolbench for machine learning and data mining. Its main strengths lie in the classification area, where all current ML approaches -- and quite a few older ones -- have been implemented within a clean, object-oriented Java class hierarchy. |
||
| Site | Version (Latest: v3.4.9) | Notes |
| ngs.rl.ac.uk | v3.4.9 | 64 bit |
License
Weka is open source software issued under the GNU General Public License.
Please see the instructions for citing Weka in your research.
Running the Weka on the NGS (General)
There a multiple ways in which jobs can be launched on the NGS
- Submission Using UI/WMS
- NGS Web Portal Submission
- Submission Using Globus
- Running a Weka "classifier" job on the NGS node at RAL:
- Running a Weka "Instances" job
This service is coming soon
After logging in to the NGS Portal, select the "Gulp" template under the " Chemistry" category and make the amendments appropriate for your specific job. Read the description page of the template for further details.
The graphical user interfaces for running Weka are not supported. Instead, Weka is run using the globus commands outlined below or from the NGS web portal. The portal has templates to stage the example data and run the following two examples. Search under the 'RAL Analysis/Stats' application repository cateogry.
We demonstrate the submission of two jobs, one using the Weka "classifiers" and the second the Weka "Instances". In the examples below 'ngsxxx' is your home directory on the target system (ngs.rl.ac.uk in this case).
To submit a job that uses the weka classifiers, executing the J48 programme and using the weather.arff data-set (located in the ngs.rl.ac.uk:$WEKAHOME/data/ , where $WEKAHOME=/apps/weka/weka-3-4-9 ). Copy the weather.arff file using gsiscp to your working directory (assumed to be "weka" in the example below.
Use the following globus command (one line):
globus-job-submit ngs.rl.ac.uk/jobmanager-lsf \
-x '& (jobtype=single) \
(arguments= weka.classifiers.trees.J48 -t weather.arff)\
(directory=/home/ngsxxx/weka)\
(stderr=/home/ngsxxx/weka/StdErr.txt)\
(stdout=/home/ngsxxx/weka/my_J48_res)' \
/usr/ngs/WEKA
This example comes from the ngs.rl.ac.uk:$WEKAHOME/data/soybean.arff file. Stage or copy to your working directory (assume to be 'weka' here)
globus-job-submit ngs.rl.ac.uk/jobmanager-lsf \
-x '& (jobtype=single) \
(arguments= weka.core.Instances soybean.arff)\
(directory=/home/ngsxxx/weka)\
(stderr=/home/ngsxxx/weka/StdErr.txt)\
(stdout=/home/ngsxxx/weka/my_vsl_res)' \
/usr/ngs/WEKA
The known good output files can be accessed at ngs.rl.ac.uk:/apps/weka/weka-3-4-9/examples
Compilation Information for ngs.rl.ac.uk
If you need information on how this application was compiled please contact the NGS support centre.
Further Information
Applications Support
The NGS cannot offer scientific support for applications. However if you require further information or believe there is something wrong with the installation, please contact the NGS support centre.
Acknowledgements
Please note: When publishing work based on use of the NGS, users should acknowledge both the authors of any programs used (see the individual program web sites, or contact the authors directly) and the NGS directly using the following line:
"The authors would like to acknowledge the use of the UK National Grid Service in carrying out this work"
This line must also accompany any use of the NGS logos.
