Tag Archive | classification tree

Classification tree tweaks

Some of you may have noticed that on Thursday we made a couple of small changes to the flow of questions that are asked for each object in Galaxy Zoo: Hubble. Both of these changes relate to the set of additional questions which we introduced during the switch from Galaxy Zoo 2 to Galaxy Zoo: Hubble. As you will have certainly noticed, the new Hubble Space Telescope images contain many more galaxies with a clumpy appearance. This type of galaxy was very rare in the Sloan Digital Sky Survey images and doesn’t really fit into the classification tree we used for Galaxy Zoo 2. To obtain useful classifications for these objects in Galaxy Zoo: Hubble we therefore decided to add another branch of questions to the “classification tree”.

Clumpy questionsDuring the first month or so of Galaxy Zoo: Hubble we have received a great deal of very useful feedback, particularly on the forum. In particular, two features of the new classification tree appeared to cause a fair bit of consternation amongst some of the Zooites. After considering your comments, and much deliberation, we decided to make a few changes.

Both points of contention related to the question asked after an answer had been clicked for ‘How many clumps are there?’. If the answer was anything except ‘one’, then we then asked ‘Do the clumps appear in a straight line, a chain, a cluster or a spiral pattern?’. Now, that’s a hard enough question to answer when there is only three clumps, but doesn’t make much sense at all when there are just two. We were trying to keep things simple but, to be perfectly honest, this wasn’t very sensible on our part. We have now changed the tree so that if the answer given is ‘two’, the question about how they are arranged is skipped.

The second issue was more interesting, because the frustration it caused told us something about the appearance of the clumpy galaxies which we hadn’t properly appreciated when planning the questions. New astrophysical insight before we’ve even collected enough clicks to start analysing! If the answer to ‘How many clumps are there?’ was ‘one’, the classification tree went back to the branch for ‘Smooth’ galaxies and asked ‘How rounded is it?’. Our thinking here was that a galaxy that was mostly just one clump would probably be an elliptical or maybe a bulge within a smooth disk galaxy.

It seems we both underestimated the discriminatory power of the Galaxy Zoo participants and how clearly different clumpy galaxies are from other types, even when there is only one clump. After having seen a few clumpy galaxies, it seems that many Zooites come to recognise that there are subtle features that set them apart from other types of galaxies. This suggests that single-clump galaxies really are a clearly different type of galaxy to the ellipticals and disks that are more common nearby. For single clump galaxies we now carry on asking the usual clumpy galaxy questions, skipping those that don’t make sense for only one clump.

Don’t worry – all your previous classifications of one (and two) clump galaxies are still safely stored away and will be very useful in helping us catalogue the subtle differences between the appearances of all these objects. Thank you, and keep clicking!