LIBSVM is a new addition to the weka3.5 later version of the feature, using this algorithm must download the jar package, configuration into the project;
LIBSVM in the use of Weka visual interface, many people have written, but the Clipse under the call material is not much, tried a lot of can not be completed, error LIBSVM classes not in CLASSPATH
LIBSVM: FQ https://www.csie.ntu.edu.tw/~cjlin/libsvm/not required
GitHub Address: HTTPS://GITHUB.COM/CJLIN1/LIBSVM
Step: 1) After downloading, unzip, find the path inside the Libsvm-3.21\libsvm-3.21\java\libsvm.jar package;
2) in Elipse Project right-click, build path to add this package to the class path path, (if the Eclipse Deployment Weka Project has a problem can see my previous blog, a detailed introduction)
3) The invocation of the interface: Classifier CFS = new LIBSVM ();
Run the file can output the results, pro-test can.
The following is a look at the Weka GUI visual interface of the call bar, one more to modify the INI configuration file process;
1 to add the Libsvm.jar Wlsvm.jar to the Weka project;
2 Then, open Runweka.ini This file in Weka's installation directory,
Put cmd_default=javaw-xmx#maxheap#-classpath "%classpath%, #weka. jar#" #mainclass #
Modified to cmd_default=javaw-xmx#maxheap#-classpath "%classpath%; #weka. jar#; Wlsvm.jar;libsvm.jar "#mainclass #
3 Select calls like other classifiers
Machine learning: The use of LIBSVM and Weka in eclipse