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.
I have a dataset of images and several ways of extracting features from
each image. For each set of extracted feature I can create an .arff file
and apply machine learning algorithm to data in it.
I would like to create few datasets (say D1, D2, D3) each with different
set of features and apply several algorithms (say A1 to A10) to each of the
datasets. It there a way to compare which combination of dataset and
algorithm gives the best performance?
I tried to add several datasets Simple experiment in Experimenter, but I
didn't find how to compare dataset-algorithm's combination performance
instead of algorithm's performance.
The only way I see that should help to solve this problem is to combine all
features for each image into a single instance and create a single dataset
and than apply several FilteredClassifiers with Remove filter, where Remove
filter selects appropriate set of features. This works, but it is a tedious
way to do it.
Is there a better way?
Thank you in advance.
Thanks for your reply (below). But how do I change the CLASSPATH
environment variable without crashing my computer? I'm using Windows
XP and have tried to see about altering the CLASSPATH environment
variable by going into start > control panel > system > advanced (tab)
> environment variables. But it looks like CLASSPATH is a system
variable and I'm not sure what I am doing. Is this where I am
supposed to add "libsvm.jar"? And am I supposed to separate it from
what is already there with a semicolon? Do I risk killing my OS?
> I'm trying to use Weka version 3.5.5 to run LibSVM. I get the error
> message "libsvm not in CLASSPATH". How do I fix this?
libsvm is a 3rd-party tool and is not included in Weka (the LibSVM
classifier is wrapper, using Reflection to call the libsvm library).
Just download it from the libsvm hompage and then include the libsvm.jar
in your CLASSPATH environment variable.
More information about libsvm:
I am samuel jebakumar doing My final Year MCA program in Karunya Univeristy
I am doing a project in weka tool , based on modification of Weka, i had
developed a new algorithm based on classifier.
To bring out novel approach in which, different methods are bisected to
generate a hybrid tree structure for tree based classification i.e. to
combine various methods like Gini index, Entropy, promoxities, information
gain, gain ratio etc.
I had added my algorithm in weka under classify tab under trees
I did some classification tasks with WEKA 3.4 using the DecisionTable classifier, and it worked perfectly without any error. Now, I'm having problems on the evaluation phase using WEKA 3.5.6 with DecisionTable. I get this error message:
Problem evaluating classifier: weka.classifiers.Evaluation
What can I do?
I suppose it's related to the classifier settings... I tried to change them, but I get always the same error message, which was not present using WEKA 3.4
I am working on some dataset in which i have 5 different date variables in
it. the date variable in the dataset is in the format of YYYY-MM-dd (ISO
8601). I can insert the dataset in to WEKA-3-7-5. But the Problem is the
format of the date variable is in "Nominal". If i want to perform any
operations on the data or with the data, i need it in "Numeric" type.
The result that i got when i applied the "ChangeDateFormat" and the Weka is
not accepting the this format and giving me the message " Problem filtering
instances: Chosen attribute not date.".
Please help me in converting the NOMINAL to NUMERIC or with the solution.
Please Reply As soon As possible.
thank you very much.
I used weka 3.5.6 to do regression on my data. I used SVMreg choosing
polyKernel, and got this:
weights (not support vectors):
- 0.3499 * (normalized) x
Can I just take this equation as the regression equation? I mean y= -
0.3499*(normailized)x + 0.5315 ? Would someone please tell me how the x is
normalized in weka?
Bioinformatics and Computational Biology program @ ISU
Ames, IA 50010
I want to do a one class classification with SMO in Weka. That is, my
training directory has only a one folder which contains the files that I
want to train the SMO with (trainDir->class1->trainingfiles). When I try
this one, it gives me the following error:
weka.classifiers.functions.SMO: Cannot handle unary class!
What I want to do is, train the classifier with only a single class and
then when at the testing phase when I fed a document to the classifier, if
it is in that class, say it or if can not be classified, gives something