Hi all I am new to Weka I have created a text file from MATLAB named temp.train.txt I am using Weka 3.5.4 / Simple CLI when i run this command java weka.core.converters.ArffLoader temp.train.txt i could see that it loads my dataset and displays on the screen. Now i want to build a decision tree for this dataset. So i used this command java weka.classifiers.trees.J48 -t temp.train.txti got the following error message : Weka exception: Can\'t open file No suitable converter found for \'temp.train.txt\'!.as i understand ArffLoader could load the data. but how to build decision tree together with Arffloader i am not gettingpls help me out.thanks in advanceArun Kumar
I am a student, working on a thesis but new to WEKA, and
can't run some experiments that I set up in
knowledge flow interface because of a memory error.
I am using weka-3-4-10 and have followed the code on the
WEKA website to change the heap size (I am
typing it into Terminal on a Mac), but below is there error I keept
getting for the code I am writing.
java -Xmx128m -classpath $CLASSPATH:weka.jar weka.gui.Main
Exception in thread "main" java.lang.NoClassDefFoundError: weka/gui/Main
Should I be entering this code in terminal? If not, where?
Can you give me the correct code for a mac and
tell me where to type it in?
In the experimenter analyze panel it can show the result of a statistical
significance test (Paired T-Tester) of the performance of different learning
algorithms, but it only shows whether a algorithm is statistically better or
worse than the others (v or *). So can we get the exact number of t and z in
Hi Weka fans,
Does anyone know about implementations of binary classifiers (i.e.,
those that are restricted to binary attributes and a binary class
which run in the Weka or Yale environments?
I have a very unbalanced training set (95% of instances are of class 0 and
5% of class 1)
and it is causing me problems to use the learning algorithms. I think it is
algorithms adjust their models based on accuracy. So they try to maximize
number of correctly classified instances. But 95% of correctly classified
~100% of TP for instances of class 0 and ~0% of TP for instances of class 1)
is already a
"very good performance" in the sense of getting maximization of accuracy.
Then the algorithm
stops. But in this case, I have a very bad model concerning the
classification for class 1.
I removed instances of class 0, such that I got a 50%-50% set, then, in fact
better. However, maybe just removing is not the best way to solve this
Now my question ( after all this bla bla bla :-) ): ***Does anybody here
know some good
literature concerning this issue of how to deal with such unbalanced
. Fabio .
I have modified the knn code(IBk.java available in weka.classifers.lazy section) available in weka to suit to my requirement.
These are the first 4 lines of my code.
I named this new file as IBk.java.
Now can any one tell me how to compile and execute this java file.
WHere to store this file?
In which directory I should be to execute this code.?
I am using weka-3.5
My main doubt is that where to place this file and compile and execute it.
Please suggest me some solution.
Thanks in Advance
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
I've updated the documentation regarding the setup of Weka within
Eclipse, since the 3.0.x HOWTO was already quite outdated.
The new HOWTO for 3.2.x is also better aligned with Weka's "build.xml"
You can find the documentation with screenshots here on the WekaWiki:
Peter Reutemann, Dept. of Computer Science, University of Waikato, NZ
http://www.cs.waikato.ac.nz/~fracpete/ Ph. +64 (7) 858-5174
I don't know whether this bug has been reported.
There is such a ARFF file:
@attribute var1 numeric
If it is opened in the Explorer of WEKA 3.4.10, it will report that
"number expected, read Token[not_num], line 6". This is correct.
If it is opened in 3.5.5, it will report "number expected, read
Token[not_num], line 2". The line number will always be "2"! This is
every inconvenient while we deal with large dataset. It is in
I am new to Weka and I wonder if it has build in stop and stem options. Also
is it possible to generate n-gram features in Weka? Or it has to be prepared
by other tools first.
Thanks a lot:)