I am using SMOreg in weka and I have to know how many support vectors were
used to make the model. Does weka give information about it? I can see the
model and count the support vectors from there, but the model is too long.
I am new to weka software and I am facing a problem in using grid search
method to optimize the parameters of the RBF kernel.
It shows the error message
"Problem evaluating classifier:"
"weka.classifiers.meta.gridsearch:cannot handle binary class !
Can anyone help me. I need it very urgently.
I'm new to WEKA software. I'm using WEKA 3.6.2 for prediction and I get
prediction score in 3 decimal points (e.g. 0.915) and when I used the same
model file and same options using the WEKA 3.4.12 I get > 10 decimal point
(e.g. 0.915467398). I'm not sure how to increase the decimal point in WEKA
3.6.2 version of the software or is there any options to get more decimal
points. This will help me to sort the samples with the high probability
score and perform lab experiments.
Thanks in advance,
Currently I am trying to use Eclipse to implement weka classifiers, say J48. I check the website, the simple implementation is
Classifier m_classifier = new J48();
/*Read Training dataset, I omit that part*/
m_classifier.buildClassifier(instancesTrain); //Here, what can I do to set up the parameter of the J48? In the weka GUI is it simple but I have no idea here.
Another question is, there are some test opinions in Weka GUI, those are:
Use train dataset //Test on the same dataset that the classifier is trained on
Supplied test set //test on a user-specified dataset
Cross-validation //perform a n-fold validation
Percentage split //Train on the percetage of the data and test on the reminder
The default setting is Cross-validation, and if I want to see the learner from "Use train dataset", how can I change my program?
I am trying to find out the relationship among three features, so I
think it would be clearer if I can visualize the three features into
one 3D graph. Does anybody has any idea about this?
Thanks in advance,
Computing and Information Sciences
Kansas State University
I have issues to utilize conversion from string to word vector.
I have test and training data set which some string data only
existing in test dataset but not training data set. As it is
applied to conversion. It results in incompatible datasets.
Since the string value will be replaced with another attribute
in test data set arff file. Any pointer to resolve the issue
for model training and validation.
WEKA accepted my multi-type data file with Numeric, Nominal, and Date attributes, and I was able to run several clustering algorithms.
I know WEKA computes an edit (Levenshtein) distance between Nominal (string) attributes, but what about Date attributes? Are these converted to seconds?
How are these Nominal/Date "distances" normalized to be commensurate with the distances between Numeric attributes?
And how do I get the output from the CLI command "weka.clusterers.Xmeans"?
I want to do resampling in weka on my data instances in java code. I know
Resample is the correct class to use. However,
I couldn't find a clear documentation (or example) on how to use the member
method createSubsampleWithoutReplacement(). So, can anybody give me an
example on how to do the resampling using the method above?