3
$\begingroup$

This is just a general question, but I ran into a few situations lately with this same framework and I'm not sure what to do with it.

Suppose we have 100 contests worth of data each with 8 individuals in it. There is just 1 individual in each of the 100 contests that can win. Suppose, I have a number of variables for each individual. What statistical tool can I use to predict the likelihood that a particular individual would win the contest?

I first thought this would be a simple logistic regression, but the fact that each row in the data set is not independent (if individual C wins, then the other 7 cannot win) seems to suggest that may not be the best tool to use here.

It effectively would be P(A wins | A has these variables and is playing against B, C, D, E, and F each with these variables).

Any ideas?

Improve this question
$\endgroup$
2
  • $\begingroup$ Maybe you could try multinomial logistic regression where Y is the winner A-F and X are the variables for each individual: en.wikipedia.org/wiki/Multinomial_logistic_regression $\endgroup$
    – user51547
    Commented Mar 29, 2018 at 18:41
  • $\begingroup$ Would I just have 100 cases for the 100 contests? Randomize the order of A through F on each case, so that Y contains all options A through F? $\endgroup$
    – Tony
    Commented Mar 30, 2018 at 0:14

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.