Extracting data characterization / meta-features for datasets

[Adriano Rivolli]

Hi for all,

I submitted 2 datasets (with and without the definition of the target values) 2 days ago.

And both are tagged as "in_preparation" and only 19 simple meta-features were computed for it.

My questions are:

1. Will OpenML generate more meta-features automatically for them? If yes, how long will it take?
2. Are there some procedure to specifically (and only) to extract the data characteristics since there is no task in the platform related with data-characterization?
3. If yes, can I characterize a dataset using only the OpenML website?

Thank you in advance.

Best regards.

Adriano

[Jan van Rijn]

Dear Adriano,

Yes, OpenML *should* generate these automatically.

I am currently working for some deadlines, but on Monday I will have time to look at this.

Please remind me if I don't get back to you.

Cheers,

Jan

--
You received this message because you are subscribed to the Google Groups "OpenML" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openml+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Joaquin Vanschoren]

Hi Adriano,

41025 has all features now

41024 has no target feature, so only the 19 simple ones can be computed.

@Jan: it looks like the 'extract_features_all' bot does run, but crashes on some malformed datasets instead of skipping them.

Cheers,

Joaquin

To unsubscribe from this group and stop receiving emails from it, send an email to openml+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "OpenML" group.

To unsubscribe from this group and stop receiving emails from it, send an email to openml+un...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--

Thank you,

Joaquin