Joaquin Vanschoren

@janvanrijn Does Fantail still compute this? It seems not, since this is empty: https://www.openml.org/a/data-qualities/DefaultAccuracy

@janvanrijn Database search confirms that this is not computed.

OK for me

@amueller Is this the result you were expecting? https://www.openml.org/search?type=data&sort=qualities.MajorityClassPercentage&order=asc

I've deleted DefaultAccuracy from the search index. @janvanrijn: shall we remove it from the database as well?

Ah, it was also in the filter options. I'll replace it.

PR submitted: https://github.com/openml/OpenML/pull/892 @janvanrijn: we still need to delete it from the database, otherwise it will come back as a data quality in the next re-indexing.

Turns out that elasticsearch has a pre-set limit of 10.000 results: https://www.elastic.co/guide/en/elasticsearch/reference/current/search-request-from-size.html and they are sorted by the evaluation measure. Hence, we only get the 10.000 'best' results, which will...

Indeed, it's currently hardcoded to 5k. Like I said, we can change the ES settings and add a button to show all if there are more than 5k runs. Lots...

Or first fetch the best results for every flow and then fill with the rest.