Closed niedakh closed 8 years ago
Hi,
I'm trying to get prediction results after testing from meka. I'm loading a prebuilt classifier: /usr/bin/java -cp "/home/niedakh/scikit/meka/meka-release-1.9.0/lib/*" meka.classifiers.multilabel.LC -W weka.classifiers.bayes.NaiveBayes -threshold 0 -verbosity 5 -t ~/engine/scikit-multilearn/meka/data/scene-test.arff -T ~/engine/scikit-multilearn/meka/data/scene-test.arff -l classifier.dump
/usr/bin/java -cp "/home/niedakh/scikit/meka/meka-release-1.9.0/lib/*" meka.classifiers.multilabel.LC -W weka.classifiers.bayes.NaiveBayes -threshold 0 -verbosity 5 -t ~/engine/scikit-multilearn/meka/data/scene-test.arff -T ~/engine/scikit-multilearn/meka/data/scene-test.arff -l classifier.dump
With verbosity 5 I am getting results such as: | 1196 [5] [0, 1, 2, 3, 4, 5]
what I get from meka.core.Result is that the first is the truth from the test set, the right are the predictions?
With verbosity 6 I am getting:
| 1196 [ 0 0 0 0 0 1 ] [ 0,0 0,0 0,0 0,0 1,0 0,0 ]
Which would suggest that with verbosity 5 meka predicts all labels, with verbosity 6 it predicts only the fifth one. Is this a bug in verbosity 5?
nvm, managed to make it work.
Hi,
I'm trying to get prediction results after testing from meka. I'm loading a prebuilt classifier:
/usr/bin/java -cp "/home/niedakh/scikit/meka/meka-release-1.9.0/lib/*" meka.classifiers.multilabel.LC -W weka.classifiers.bayes.NaiveBayes -threshold 0 -verbosity 5 -t ~/engine/scikit-multilearn/meka/data/scene-test.arff -T ~/engine/scikit-multilearn/meka/data/scene-test.arff -l classifier.dump
With verbosity 5 I am getting results such as: | 1196 [5] [0, 1, 2, 3, 4, 5]
what I get from meka.core.Result is that the first is the truth from the test set, the right are the predictions?
With verbosity 6 I am getting:
| 1196 [ 0 0 0 0 0 1 ] [ 0,0 0,0 0,0 0,0 1,0 0,0 ]
Which would suggest that with verbosity 5 meka predicts all labels, with verbosity 6 it predicts only the fifth one. Is this a bug in verbosity 5?