Testing clustering algorithms for accuracy in weka Experimenter

Question

I'm using Experimenter to compare clustering algorithms to test accuracy. I couldn't find clustering algorithms in the Simple experiment mode, so I used Advanced and did the clustering using these steps:

https://waikato.github.io/weka-wiki/experimenter/running_an_experiment_using_clusterers/

These are my settings:

And this is my analysis screen.

I can't find 'Percent_correct' here though to test accuracy. How can I test for accuracy?

Thank you.

fracpete · Accepted Answer · 2023-12-02 19:01:26Z

1

You can't. Accuracy (ie Percent_correct) is only available from supervised algorithms like classifiers and cluster algorithms are unsupervised.

answered Dec 2, 2023 at 19:01

$fracpete's user avatar$

fracpete

2,6082 gold badges14 silver badges21 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

not_castillo Over a year ago

I see, thanks! Is there a way to compare the % of incorrectly clustered instances though, like in Explorer?

fracpete Over a year ago

Unfortunately, no. You'd have to write custom code for obtaining the classes-to-clusters information and post-process that further.

Collectives™ on Stack Overflow

Testing clustering algorithms for accuracy in weka Experimenter

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related