Come up with a new, catchy tag

Question

Come up with a new, catchy tag

Closed this issue 7 years ago · 18 comments

mfeurer commented 7 years ago

to replace OpenML100

Answer 1 · 2017-11-10T16:00:12.000Z

please also note that the new suite will have only 60-80 sets

Answer 2 · 2017-11-10T16:06:31.000Z

OpenML2017?

On Fri, 10 Nov 2017 at 17:00 Bernd Bischl ***@***.***> wrote: please also note that the new suite will have only 60-80 sets — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#7 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABpQV5ecNq-83mr6otJGoGUhoGsHezCtks5s1HMMgaJpZM4QZvkd> .

-- Thank you, Joaquin

Answer 3 · 2017-12-22T08:48:52.000Z

HighQualityData2018?

Answer 4 · 2018-01-16T13:00:47.000Z

I checked the old paper, it refers to the following hyperlink:
https://www.openml.org/s/14

We could set it up in such a way that this link refers to the 'old' openml-100 datasets, but tag them with the tag 'OpenML100-depricated' or something like that. That would allow us to reuse the tag 'OpenML100' for the new study.

We are all using this term anyway, it would be hard to change it here in the lab.

Answer 5 · 2018-01-30T16:52:51.000Z

According to the google doc there are 81 datasets (minus some potential candidates for removal), so we really shouldn't stick to a number. How about creating a nice abbreviation: OpenML Small Binary Classification Benchmark Suite, short OML-SBC18 benchmark suite. It's not as nice as OpenML100, but contains all necessary information and is easy to remember.

Answer 6 · 2018-01-30T17:03:19.000Z

They are not all binary? And not all small? The latter is of course subjective :)

On Tue, 30 Jan 2018 at 17:53, Matthias Feurer ***@***.***> wrote: According to the google doc there are 81 datasets (minus some potential candidates for removal), so we really shouldn't stick to a number. How about creating a nice abbreviation: *O*pen*ML* *S*mall *B*inary *C*lassification Benchmark Suite, short OML-SBC18 benchmark suite. It's not as nice as OpenML100, but contains all necessary information and is easy to remember. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#7 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABpQVzcbwBVlzg3xEtEe8IZlCSTknWtwks5tP0jkgaJpZM4QZvkd> .

-- Thank you, Joaquin

Answer 7 · 2018-01-30T17:06:24.000Z

How could I add the word binary in there??? Well, what about:

OML-SASC18 (simple and small classification)
OML-SC18 (small classification)
OML-CBS18 (classification benchmark suite)

Answer 8 · 2018-02-01T16:02:49.000Z

@giuseppec and @joaquinvanschoren and I decided that we should cast a vote about the name, I do suggest to use a voting procedure like Condorcet with Schulze tie breaking as implemented here: http://www1.cse.wustl.edu/~legrand/rbvote/calc.html

Answer 9 · 2018-02-06T20:39:55.000Z

OpenML-Light-2018 (And we can use OpenML-Heavy for large datasets ;))

Answer 10 · 2018-02-07T16:47:58.000Z

That name does not highlight that we do classification.

Answer 11 · 2018-02-07T17:18:08.000Z

OpenML C-Scape18?

-scape means 'a wide view', which is what a benchmark should also offer :)
And Seascape is a nice word :)

We can use R-Scape for regression, M-Scape for multi-label, I-Scape for imbalanced,... :)

Answer 12 · 2018-02-23T13:57:07.000Z

OpenML2018? Small and catchy

Answer 13 · 2018-02-23T14:06:51.000Z

CSC18 - Curated Small Classification 2018

Answer 14 · 2018-02-23T14:08:38.000Z

OML-CC18 or OpenML-CC18

Answer 15 · 2018-02-23T14:08:56.000Z

OpenML-CC18 ?

Answer 16 · 2018-02-23T15:49:30.000Z

did we decide something?

Answer 17 · 2018-02-23T15:54:13.000Z

Unless anyone has a very strong different opinion, I propose it's going to be OpenML-CC18

Answer 18 · 2018-03-15T15:08:11.000Z

Closed per Skype call (@frank-hutter @giuseppec @mfeurer @janvanrijn) as no one complained about OpenML-CC18