Tuesday, February 3, 2015

Base R Assessment!

Test your skills with this R-powered R assessment of base R knowledge!

Built using the R powered adaptive testing platform Concerto, this assessment provides a short but powerful tool at evaluating your base R understanding relative to that of your peers.

Currently the assessment has over seventy items (questions) in the pool while each individual takes less than twenty of these selected randomly. So each test is unique.

Those who score well enough will be given the chance to contribute their own items to challenge other users.
 
http://concerto4.e-psychometrics.com/?wid=13&tid=14
http://concerto4.e-psychometrics.com/?wid=13&tid=14

10 comments:

  1. Fun idea, but the test is rather slow. This is not surprising, since the test is R-powererd. Almost ironic.

    ReplyDelete
    Replies
    1. Good point about speed. Entirely a server side issue rather than an R specific issue though of course there would be less demands on the server if R worked more efficiently.

      Delete
    2. Looks like the issue was with a faulty javascript slowing down the server.

      Delete
  2. I started taking the questionnaire and got about 4-5 questions deep, however the time it takes to generate the next question really slowed down dramatically -- I was left looking at a little spinning wheel at the top of the browser for what felt like an eternity, and just gave up.

    It looks like fun, though. I'd like to run it all the way through without investing an entire hour to do so, so hopefully you folks can fix that!

    ReplyDelete
    Replies
    1. Thanks, sorry about that. I am just using the Concerto pilot server which seems to have been taxed by this task.

      Delete
  3. @Francis, very cool. it's possible to get multidimensional CAT running on shiny with mirtCAT http://cran.r-project.org/web/packages/mirtCAT/index.html .

    ReplyDelete
    Replies
    1. Thanks for the suggestion Ricardo. I plan on investing some energy into learning about mirtCAT. However, my initial impressions were that it was pretty buggy.

      Delete
    2. I'd be interesting in hearing what bugs you've found in mirtCAT. Feel free to open an issue on Github if you do. Cheers.

      Delete
    3. Hi Phil, thanks for this very reasonable suggestion. I plan on investing more energy in leaning mirtCAT and I will make sure to bring up specific issues as they arise. Sorry if this seemed like a low blow. I was just using getting acquainted using the vignette and there is an example CAT with addition items. Strangely I was working through it and one of the automated generated items did not have an answer that was correct. Since the item was something like 4535+3242 it was easy to verify that the answer was incorrect. Some of my previous work with Shiny made me feel that this type of situation was not uncommon (one of the reasons I declined from writing shiny adaptive testing software). But, maybe it was just a fluke.

      I am also sorry if I came off as too harsh in any ways. You have done a great job with the package. I am personally looking forward to working with it into the future!

      Delete
  4. Nice idea, but needs refinement.
    This test will have big difficulties to differ inbetween a novice person and a experienced person.
    buggy and it hangs

    ReplyDelete