From WEKA 3.7.2 installation and use of LibSVM in WEKA has been made easier by the creation of a LibSVM bundle that can end up being installed making use of either the graphical or command word line package manager.The subsequent steps are usually for Home windows XP (unfortunately, the GUI changes among the different Windows variations).This can take place where the CLASSPATH will not get extended to its real worth in starting up WEKA.
You can inspect your current CLASSPATH with which WEKA got began up with the SimpleCLI (notice earlier bullet point). If CLASSPATH will be listed presently there, your program offers the exact same problem. We made a quick fix including an stationary Random attribute to libsvm.svm course. Therefore if there are usually classes in the datasét with zero occurrences through all the situations, LibSVM feels that these courses dont exist whereas WEKA knows they can be found. The dimension of the selection is equal to the number of lessons ( 22). On the additional hands, if this technique will be invoked from Iibsvm.svmpredict, the course that certainly not appears is certainly ignored, therefore the range dimension can be now equal to 21. I believe that much better results are usually attained if lessons without instances are overlooked, but I dont understand if it is certainly very fair. In reality, accuracies from wéka.libsvm and fróm libsvm.predictsvm seem to be the same if the course that under no circumstances appears will be eliminated from ARFF file. Moreover, probably the mismatch bétween the training amount of classes and the testing amount of courses is certainly the cause behind worse precision outcomes when svmpredictprobability invocation will be produced fróm WEKA, but I havént demonstrated it yet.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |