A first look at the data from Galaxy Zoo 2 beta
“Wow! Just wow!”
Over the last week I’ve been taking a look at the data that many of you have been generating on the Galaxy Zoo 2 beta site. It’s great to see that the Galaxy Zoo magic is working again, and this time we have so much more information about each galaxy! We really are going to be able to do some amazing research with the Galaxy Zoo 2 data.
However, the current task is quality control, in particular making sure that the answers we are getting correspond to what we had in mind when we came up with the Galaxy Zoo 2 questions. We want to be sure that we aren’t asking questions that can’t be answered given the typical quality of SDSS images, or that aren’t going to give us useful information about the galaxies. We also want to see where we may be able to improve the questions to better discriminate between different types of galaxy, and where we should rewrite the tutorial to give more guidance.
This testing was one of the reasons for having a beta stage. However, even though the data collected so far was for mainly for quality control, we will definitely also be using it for science. In particular, it will give us a preliminary idea of the results we can expect to find in the full Galaxy Zoo 2 dataset, which we can advertise at conferences this spring. Thanks to all our beta testers for helping us to ensure that Galaxy Zoo 2 will be a success.
To show you how well the current Galaxy Zoo 2 system is working, here is a prettified version of the page I made which got me saying “Wow!”, Chris jumping around his office, and made the whole team very happy when it was sent around on Friday. (Sorry if the images load slowly – there are a quite a lot of them!)
I’m now adding more quantitative information for the team, such as graphs of the distribution of answers for each question, to test all the things I described above so that we can improve the tutorial and classification system ready for the imminent full launch of Galaxy Zoo 2 to the world.
Galaxy Zoo 
Very impressive! Hey Alice–you can start making slides for your next talk! 😉
LOL Georgia, what talk is this? 😀 CONGRATULATIONS zoo team, you must be so excited!
Helpful demonstration of both complexity and clarity. The ultimate tutorial from a clicker’s perspective. Do we know whether we will continue with this project?
Really a great collection.
Thanks for update
WOW!!!! JUST WOW!!!!!! this is SO EXCITING!!!! Thanks for the feedback – Zoo2 is fun to do and is never boring because of the variety of galaxies that appear and the urge to do just one more to see what the next one will be like!
WOW indeed !
Go GZ2 !
Great news, thanks for the update Steven! 😉
Thanks, Steven, it’s really good to see it all set out like this, very clearly defined answers. The hard work is paying off, I’m sure.
Regards
Thomas
It is great to hear that the ad hock stage of GZ 2 works so well. Thanks, Steven for the update. Congratulations, TEAM for a good start for the Internal Year of Astronomy!
This article has been added to the Astronomy Link List
Splendid to see. As Jim said, this makes a good tutorial. Congratulations and thanks for taking the trouble.
Great pictures to confirm selection criteria.
WOW! is right! after looking at the images I was saying wow too. I agree that these images should be in the tutorial so people can continually revert back the the page to see what kind of images should be classified in what way.. I know going back to the tutorial on GZ1 helped me recalibrate.
I looked at Astronomy Link List with great interest, but couldn’t work out precisely what it was supposed to be other than a list of each day’s updates on various websites – can someone give me a more detailed description than that? Is it good?
Mark, I’m working on refining the GZ2 tutorial right now!
I have a 2-3 disagreements with the beta examples page that mainly have to do with the apparent number of spiral arms, but for the most part it’s good. It does have a lot to do with the grainy mediocre quality of most of the data. Also, because of that, we aren’t really taking into account any faint lenticular features of supposed “elliptical” galaxies. I tend to see these faint lenticular features amidst the graininess where others only see an elliptical galaxy. Don’t get me wrong though, the beta is an improvement over Galaxy Zoo 1.
You must bear in mind that the ‘grainy mediocre quality data’ is from one of the biggest and best astronomical surveys to date! Better quality images = less area of the sky covered = fewer objects big enough to be able to classify. We’re working at the edges of what is possible to try and generate ground=breaking science. Lenticulars are very hard, even for experts. We are trying to work out how to get them classified better by Galaxy Zoo. For now, remember that lenticulars are disks with no spiral features, so classify them like that and we’ll get them – and try to statistically account for those that are missed.
I have learned that gravitational lensing events are hard to come by. It would be interesting to hear some professionals’ opinions on the pictures in the examples classified as “lensing”: which of them might truly show lensing effects.
I suggest you throw in a self-test image periodically (identified as such) where we choose our settings, and then get a comparison with the canonical answer (and explanations). The name of the game with QC is feedback. I know I’d like as much as I can get.
Also, is anyone else having the experience where, in maybe 1 out of 5 cases where I’m asked to compare two galaxies (more prominent bar/spirals) the right-hand image doesn’t appear?
I had that happen a couple of times today, dkretz.
The comparison images seem to be a random selection. Would it not be better to use a ‘standard’ set of images for comparison purposes? Then you know whether the image under question is less or more like these standard ones. Maybe you already do this… although it doesnt appear like it.
Greetings Fellow Zooicons!
I was reading the Feb 8th comments above on “grainy quality” and lenticular characteristics of elliptical galaxies. I too have alot of grainy quality issues (possibly due in part to what’s left of my eyes- may have to dump the Walmart reading glasses!). Anyway, having the option in Galaxy Zoo 2 to classify a Galaxy as a Lenticular (that is a disc with no obvious spiral features) would solve many of my challenges trying to differentiate between some spirals and ellipticals. Will there be an option classify a galaxy as a Lenticular in Zoo 2?
Dear Colleagues,
I cannot see a consistent measureable difference between the spirals in the beta examples page with “Dominant” bulges and those with “Obvious” bulges. Is there a rule or better way to quantify?
i have always been interested in the universe and now that i am taking part in the galaxy survey i am even more so. keep up the good work
Steven, I’m not criticizing the Sloan Survey Data. It’s the best we have to work with. I’m just saying there should be an extra option for the “ellipticals/Smooth” that appear to have lenticular features (the Smooth category should have a few more options anyways). Don’t forget Lenticulars are categorized as S0, not E0. They represent the transition between Spirals and Ellipticals, and a some Lenticulars DO have faint dark dust spiral or ring features.
WOW Its great that what we’re doing for our pleasure is helpping towards scientific data gathering and analysis. I loved GZ1 and have really enjoyed doing the Beta GZ2…… And I agree with Gendo there really should be an option for non-spiral non-elliptical galaxies.
Very impressive data! This clearly works and works very well. It great to see some much input from people like myself who just enjoy discovering and helping out the scientific community. My congrats to everyone. :0)
Though I think I’m classifying OK, the scentist in me does not trust me. I hope you’re sending out each galaxy at least twice and throwing mismatches back in the pit until there’s a majority result, or an escalation to an expert!
p.s. now I know how BOINC feels. 8^)
Re complexity of classification. Its a tricky one. I have this T Shirt (http://www.nhm.ac.uk/nature-online/british-natural-history/survey-bluebells/index.html) – new survey not live yet for 2009!
We found that the number of advanced/ expert contributors was so low that data wasn’t really usable. I’d /suggest/ having a free-text area for notes when we feel there’s something outside your decision tree or ‘other’ isn’t useful enough.
mike – The idea is for many people to classify each galaxy, providing an excellent error check: “the wisdom of crowds”. In GZ1 we ended up with around 40 independent classifications per galaxy! We carefully analyse and correct the results for biases before using them for science, and compare with a subsample of expert classifications. We plan to introduce a ‘tagging’ feature in the near future to allow for information outside of our question’s scope.
The bluebells survey sounds very interesting – good luck with it.
Gendo and Louise – believe me, I know about lenticulars, and I am very interested in working out how they fit into our picture of galaxy evolution. For those reading this who don’t know, a lenticular is a disk galaxy, usually with a big bulge, which does not display spiral arms. If you think a galaxy is a lenticular, then you are judging it to have some sign of a disk, so you should click “features or disk”, not “smooth”. Then you should answer the following questions appropriately, e.g. no spiral arms. Maybe I should make this clearer in the tutorial. We are hoping to identify lenticulars by some combination of disks with no spirals and smooth, cigar-shaped objects, and attempt to statistically account for the fact that face-on lenticulars are very hard to differentiate from ellipticals.
Looking good and looking forward to getting involved 🙂
I hope the ID #’s will eventually be each image. Without them we will not be able to dicuss with others on this website what we have seen. Also,on GZ1 we could back-pedal to double check our work. I know we can on GZ2 to a point. But if you finish one and had second thoughts immediately after, you could hit the back button and look at that object again. Now, on GZ2, that is not possible. Anyway, just those minor things, overall….I’m really very happy. We are wringing many more times the info now than before. Very satisfying
Thanks guys : )
RobertGB
Hi.
When logging on to GZ2 my saved password is never remembered! I have always to go through the process of new password creation every time I try to log on.I have done this with other people checking while I do this, but they agree I don’t seem to be doing anything wrong. Is there a problem with the log on process?
Regards,
Walter rowan
I’ve just been introduced to Galaxy zoo 2 a few day ago
and just love it!. But I have can’t figure out how to
bring up the galaxies I’ve added to “my favorites” and
don’t see any identifying numbers on my images of the
galaxys. I’d like to post some with comments of what I’ve observed. How do I do this. Love what I do though. Please have patience with this old lady with
computers. She was a total luddite a few years ago.
Love it!
Gina W
Steven said, “the wisdom of crowds.” I call it “mob knowledge.”
I agree with Gina W. I miss the numbers for the very strange and non-galactic oddities which occasionally turn up. This is how Hanni’s verwoerp came to light. Most of my queries in GZ1 images are catered for in GZ2, though.
Age means that my eyes are not too good either. I have found Magnifixer, when you get used to it, and when treated gently, is a very good freeware screen magnifier for GZ1 and 2 (and other) purposes. Can I also recommend Gadwin PrintScreen (freeware too) to anybody interested.
Nice to have the opportunity to do something constructive online.
hello sounds interesting hi jazzyamy123!!!!
I just joined the group yesterday. What an amazing site! Unfortunately I have a picture of the ‘Hubble Deep Field North’ as my desktop, and I’m feeling the need to analyze all the galaxies!
What I’d like is a “What the HECK” button for when it is too grainy and too distant to really know what it is.
help i am afraid i have lost the plot,