Galaxy Zoo Starburst Talk

Overlapping histograms in Tools?

  • JeanTate by JeanTate

    One of the first analyses I'm going to do - unless someone beats me to it! 😃 - is various 'overlapping histograms', which is a histogram plot containing the (binned) distribution of two datasets, by the same variable. For example, the Quench Sample (QS) and Quench Control (QC) by Abs R (K-corrected, de-redened r-band absolute magnitude) ... QC has been chosen to match QS in redshift and stellar mass, so differences in the Abs R distributions may point to something interesting (actually, there are likely much more interesting plots, but this one is easy to visualize).

    I can create two separate histograms easily enough, using Tools, but do not know how to combine them.

    Anyone?

    Posted

  • Nax by Nax

    I was gonna ask about the exact same thing. It would be very handy to compare the two datasets. Wanted to try to explore any relationships with galaxies with similar redshifts in each dataset, and various other things. Doesn't seem possible now, maybe the ability can be brought in?

    Posted

  • JeanTate by JeanTate

    Here are two histograms of binned Petro_R50, one is QS (filter Petro_R50 <10), the other QC (same filter)1. Unfortunately, I forget which is which, and don't know how to find out!

    http://tools.zooniverse.org/#/dashboards/galaxy_zoo_starburst/51f2d66c0aab2a03e20000db

    http://tools.zooniverse.org/#/dashboards/galaxy_zoo_starburst/51f2d5120aab2a03ef0000e0

    Of course, histograms are simply nice visuals; for the paper we'd need to do a statistical test, to check how significant the apparent difference in distributions is (perhaps a KS testWP?). Even if we could get the two sets of data together, how would we do such a test, using Tools?

    1 Yes, it is important to check out the apparent outliers, with objects with Petro_R50 > 10.0". But this is just a test ...

    Posted

  • wassock by wassock moderator

    Jean when I follows the link I see just the empty windows on the dashboard, I dont seem to have the data?

    Posted

  • JeanTate by JeanTate in response to wassock's comment.

    Yeah, me too.

    I don't know what's happening, I followed the steps in Sharing Tools Dashboards, and checked that both links showed what I expected (more or less) after I'd posted them. Ah well, back to using the more time-consuming and cumbersome - but more certain - method of taking screen shots and posting them ...

    Posted

  • edpaget by edpaget in response to JeanTate's comment.

    Man I mystified by this bug too. Please don't delete the dashboard so I have a chance to figure out what's happening!

    Posted

  • JeanTate by JeanTate in response to edpaget's comment.

    May be too late; I think I may have - accidentally - deleted one of them (but the other should still be there).

    Posted

  • edpaget by edpaget in response to JeanTate's comment.

    Hey Jean, it looks what happened is that you linked a tool to its parent tool, which caused the tools to keep calling each other until the browser runs out of memory. I'm going to put in a check to prevent you from being able to create circular links. I can probably get that dashboard fixed although I don't have direct access to our database, so I'm not sure when it'll happen.

    Posted

  • JeanTate by JeanTate in response to edpaget's comment.

    Thanks Ed. I don't really understand exactly what "linking a tool to its parent tool" means, and have no idea how I could have done that anyway! I suspect, however, that it may not be uncommon, and may explain several of the problems others have reported (in various threads and posts) ...

    Posted

  • edpaget by edpaget in response to JeanTate's comment.

    So you had a Histogram connected to a Table, but that Table was also connected to the same Histogram. The Table then asks the Histogram for a piece of information, which causes the Histogram to then ask the Table for that piece of information, and the cycle continues until the browser crashes.

    The Dashboard should prevent you from doing that now.

    Posted

  • JeanTate by JeanTate in response to edpaget's comment.

    Thanks again Ed.

    Good to know I can't do that again.

    I'm still puzzled; how did I manage to get myself into such a loop in the first place? Ah well, maybe I'll never know ...

    ETA: Should I delete my existing dashboards? Or will those with this kind of problem give me an error message as soon as I touch them?

    Posted

  • edpaget by edpaget in response to JeanTate's comment.

    No, your existing dashboards should be fine. If another one fails to open like that one, post here or email me at ed at zooniverse dot org and I'll see what's causing the problem.

    Posted

  • JeanTate by JeanTate

    Thanks.

    How about this one (source)?

    http://tools.zooniverse.org/#/dashboards/galaxy_zoo_starburst/51fd0ef4be70a677d2000007

    It comes up with a blank/empty Table, ditto Scatterplot.

    Posted

  • edpaget by edpaget

    It just looks like a set of data wasn't chosen. http://tools.zooniverse.org/#/dashboards/galaxy_zoo_starburst/52025e890aab2a14120001b4

    Posted

  • JeanTate by JeanTate in response to edpaget's comment.

    Thanks.

    The data most certainly was chosen, because when I'd finished I had a very nice plot ... but when I went back to it later, nothing.

    Anyway, if I want any of these, I'll re-do them, from scratch.

    Posted