Weka

Many of the fuzzy-rough feature selection measures have been ported to Weka – the latest standalone program (i.e. the runnable jar) is here, updated June 2023. (Old version Feb 2021 is here.) This includes several non-parametric statistical tests in the Experimenter: Mann-Whitney U test, Wilcoxon signed ranks test, and Friedman test with Holm’s post hoc procedure.

If you want to process larger datasets, then you’ll need to change the Java heap size. An easy way of doing this is to put this file: run.bat in the same directory as Weka (above) and double-click it. You can edit this file to increase or decrease the heap size.

Many of the algorithms can be found in KEEL and some have also been ported to Python and R.

A worksheet that explains how to use Weka and several of the implemented fuzzy-rough algorithms can be found here.

FuzzyRoughSubsetEval contains various measures and the option of selecting connectives, similarity measures etc. HillClimber is similar to Weka’s Greedy Stepwise search, but allows search to a degree alpha (for alpha-decision reducts) and search using non-monotonic measures (such as VQRS). AntSearch implements ACO-based search and includes the option of using any subset evaluator for the initial heuristic matrix and subset evaluation. PSOSearch implements search based on binary particle swarm optimization.

In the Classifiers section, there is now a range of fuzzy-rough classifiers including fuzzy-rough nearest neighbours and QuickRules. There are also several filters for fuzzy-rough instance selection.

28 Comments on “Weka

  1. It is really a commendable work. I really like people contributing their algos to WEKA. In no time Weka will emerge has a standard platform for data mining.

  2. I was looking for something which can be used on macbook! anyway your version beautifully functioned on my pc! thanks again 🙂

  3. I have put the batch file in the same directory where i have downloaded weka. When I double clicked batch file, a blinking black screen appeared and disappeared.

    • Sorry, then, I don’t know what the problem is. If other executable jar files work then I’m not sure what’s going wrong. If other executables don’t work then you could try reinstalling the JRE.

  4. Hi
    How should I add and write a program in weka? for example use weka in netbean
    I want to mix genetic algorithm and Iaco as feature selection and add it to weka
    Do you have a refrence to propose me?
    thanks

  5. I used your weka.jar file. if I take the text file other than english there is not having any option for utf8 as in the original weka have. could you tell me how I used utf8 in your WEKA jar file

  6. hello sir,
    i have working in weka version 3.7.2
    i do PSOsearch…
    i need to setup fitness function…how to setup fitness function in weka?
    thank lot sir

  7. Thank you so much for your site, it has really been helpful to me. I am using the WEKA version of your tool. In WEKA filters, and supervised instances, you have a number of filters for preprocessing, i was wondering if you have written anything about the following filters;

    a)MinorityObjectGenerator
    b)StratifiedRemoveFolds
    c)SMOTE

    or if you can give me links to where i can learn about how they work theoritically .

    Kind Regards,

  8. Dear Dr. Jenson,
    This query is related to the rough fuzzy based WEKA tool available from your website (http://users.aber.ac.uk/rkj/?page_id=139) and working with noisy data. You have added two filter to add noise to a dataset. Would it be possible to provide some more details (or even the publications discussing the details of these two noises)) about both ” weka.filters.unsupervised.attribute.AddConditionalNoise” and “weka.filters.supervised.attribute.AddClassNoise”

    thanks,

  9. Hi admin I did all things that said here to run the fuzzy rough set reduction but I don’t have that option on my weak?
    please help me I need it in an emergency.
    thanks

  10. Thanks Mr.Jensen, I really don’t know much about data and weka. And your folk version is really help me.
    Best wishes for you.

Leave a Reply

Your email address will not be published. Required fields are marked *

*