I am looking at code from HMMWeka and noticed the following.
On isNominal() data e.g. when attr.relation().attribute(0) returns
a call to attr.relation.numDistinctValues(0) returns 0 instead of 2.
What am I missing here?
> On Jun 2, 2017, at 2:15 PM, Alexander Osherenko <osherenko(a)gmx.de> wrote:
> I tried to find the problem and didn't. Maybe you remember I have Weka
> 3-8-1 and AutoWEKA installed and they are running without any problems on
> XP. That's why I have to seek further
> he difference is: Weka runs only on XP successfully both from the Windows
> menu and console (java -jar weka.jar); on Win10 only from the console.
> I’m sorry I missed that the discussion had switched from AutoWeka not
> working to Weka itself on Win 10 not working?
The problem is: I installed java, update 80 and Weka 3.8.1 on two OS: XP
and Win10. Although the installations should be identical, something is
wrong. On XP, the installation works perfectly: I can start Weka from
Windows menu and from the console. There are problems on Win10 -- I can
only start Weka from the console (java -jar weka.jar), but not from the
> Just an idea. I found in the src of the Auto-WEKA package a method that
> could provide a hint of differences.
> Changing the java.home property might help Auto-Weka but it shouldn’t help
> Weka Explorer launch.
> Maybe there is something in Weka log concerning why the application isn’t
> launching. That should be at…
> For whatever user.home is for you on Windows.
There are no entries in this file concerning any errors. I don't think
that Weka has the possibility to take control over execution since startup
is aborted before, for example, because the weka.jar is not found.
> Let us please compare the settings in Weka Systeminfo. I found some
> differences, for example, in java.home and sun.java.command. Can I set
> these variables in RunWeka.bat?
> How are you comparing this for Win 10 when the application doesn’t launch?
> Again the java.home property shouldn’t help the Explorer application itself
> launch. I don’t know what the sun.java.command is, I don’t show it set for
> my own application on OS X.
I can start Weka on Win10 only from the console
(java -jar weka.jar)
that starts in Weka GUI chooser
I can choose the Help-Systeminfo command that shows
a windows with
, for example
, sun.java,command (
[image: Inline-Bild 4]
on Win10 the value of sun.java.command is weka.jar; on XP the value is
weka.gui.GuiExplorer (or something like that).
I am working with unbalanced datasets, similar to the diabetes dataset (500
tested-negative and 268 tested-positive, total instances: 768) .
I wish to measure how much the Precision on the minority class of a Naive
Bayes classifier (standard parameters) *improves* wrt the Precision of a
random classifier on the same class.
Since the dataset is unbalanced, it does not make sense to use ZeroR, since
ZeroR only guesses the majority class.
Since I thought that is NOT theoretically sound to compare an evaluation
measure like Precision (TP/TP+FP) to the probability distribution of the
individual classes (ie. total instances of a class OVER the whole
population), I was suggested to use a random classifier, or better a *weighted
guess classifier, *as described in these pages:
Making the calculations (hopefully they are correct, but I cannot swear on
them :-) ), it turns out that the Precision of the *weighted guess
classifier *always corresponds to the probability distribution of the
individual classes. That is, in the case on the diabetes dataset, the
distribution probability of the minority class (ie the "tested-positive")
is 0.348 which is identical (according to my calculations) to the
Precision on that class returned by the *weighted guess classifier*.
The Naive Bayes' precision on the "tested-positive" class is 0,678.
Can I safely (and soundly) say that the NB improves 0.33 points on the
precision of a random classifier?
Do you know if there exists any implementations of a *weighted guess
classifier*? Making all the calculation is tedious...
It would be handy to have it available in the weka workbench and use it as
Thanks in advance for your answer.
Can someone clarify me what's the Java platform version on which Weka 3.8
runs? Can it run on any (for example, the latest) version or is it for
example restricted to the particular version on which the software was
Hi Weka experts,
I have designed a classification ensemble technique, the algorithm has been
presented in a conference and published in conference proceedings. Further,
the algorithm has been tweaked to handle imbalanced datasets and has been
extensively tested on various imbalanced datasets.
Recently I have implemented both versions of algorithm by integrating two
new classes in WEKA. My question is : How do I get these algorithms
published as a part of WEKA, when the new version is released?
Department of Computer science