My friend gave me this puzzle and I just can't see a way to approach it... (I would like to use SQL)
Let's say I sit a test at school, 20 questions, and get I 10 answers right and 10 wrong.
The teacher wants to compare my answers to the rest of the class.
They start by finding the other students who got the same questions correct (ie at least the same 10 right answers).
Then we want to examine the other correct answers achieved by that group and find (in order) the questions that were most often answered correctly by the group. So if Bill Fred and Mary in my class all got the same 10 answers as me right, but also got Quest 16 correct (which I got wrong) maybe I should go and bone up on Q16 as it might be easy for me to learn that topic.
Could someone point me in the right direction. I keep getting lost when I think about creating joins because the fields would be so dynamic (I could do it for a small set, but what if there are 5000 possible questions and we have all been asked different subsets over hundreds of classes etc etc). This would be something I would like to do on the fly in a program and I guess has application to all sorts of forecasting - but I'm not a stats person so feeling a bit lost).
Have a great weekend!!