yes i have used this command in command line
java -cp moa.jar -javaagent:sizeofag.jar moa.gui.GUI
i have got some error regarding this
Exception in thread "main" java.lang.ClassNotFoundException:
at java.security.AccessController.doPrivileged(Native Method)
FATAL ERROR in native method: processing of -javaagent failed
so please can u uhelp me regarding this problem
please sir it is neccessary
Dear WEKA community,
I am not an experience WEKA user, so I need some information from you.
I am carrying out an investigation on the k-NN algorithm and its variations.
And I am experiencing the following problem.
Distance weighted voting with the k-NN classifier uses the formula
1/distance. However, what will happen when distance = 0 ? This scenario
implies full similarity between a training instance and a test instane. In
the literature, it is said that the classification decision in such cases
should occur only on the basis of the dominant instance(s) [distance = 0]. I
have implimeneted a sample tool, which realizes the the algorithm as in the
literature, but I get different results with the same data sets when using
WEKA and my tool. So my question is, can anybody share his/her experience
with me and tell me how is WEKA working in this case ? How will the case 1/0
be treathed by WEKA. I tried to search to the source code, but with no
Thank you in advance!
View this message in context: http://old.nabble.com/Ibk-%28K-NN%29---how-is-1-distnace-handled-when-distn…
Sent from the WEKA mailing list archive at Nabble.com.