Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Multi label classification

Hi,

Does RapidMiner support Multi-label classification? We have a use case in which we have a set of feature columns and multiple target columns. Each target column will be a classification label.
This is different from multi-class classification with one target column and multiple classes.

In auto-model I saw that we can only select one column with role as 'label'.

Regards,
Divya

Find more posts tagged with

AI Studio

Classification

Accepted answers

tftemme

Hi,

Currently RapidMiner does not support Multi-label classification easily. You could use Loop over your label attributes, set the current to Label role, train a model on it and collect the models. This has to be done of course in the process designer, as it is a too specific setup for AutoModel.

With 9.4.0 we will release a Multi Label Model which is capable of exactly doing this 'looping' over the label attributes in more convenient way. See also this thread: https://community.rapidminer.com/discussion/55735/set-two-featurs-as-label-at-the-same-time for more details.

Hopes this helps,
Best regards,
Fabian

All comments

tftemme

divya_das

Thank you Fabian.

MarlaBot

Hi @divya_das - this is MarlaBot. I found these great videos on our RapidMiner Academy that you may find helpful:

Instructional Video: Automatic Classification of Documents (Viewing time: ~9m)

Instructional Video: workshop: Regression & Classification (Viewing time: ~1m)

Instructional Video: Auto Model - Classification (Viewing time: ~9m)

Please LIKE my comment if it helps! 👇

MarlaBot

SGolbert

Nice work on the MarlaBot, it should become unicorn soon!

sgenzer

LOL @SGolbert actually I cringe half the time I see MarlaBot's posts. Of course right now it's doing nothing but creating a training set

#dogfooding

Scott

SGolbert

I would feel better if it would respond only if after some days no one has bothered to answer (or perhaps the post was too vague or too broad)

tftemme

To update on this. As stated, since 9.4.1 RM contains now the 'Multi Label Modeling' operator to train a model for multiple label attributes. There are also the operators 'Multi Label Performance' (for evaluating the performance of such a model), and the 'Multi Horizon Forecast' and 'Multi Horizon Performance' operator which handles the time series related multi label operations.

Best regards,
Fabian

divya_das

Thank you Fabian.