Search the Community
Showing results for tags 'regression'.
Found 1 result
Making the assumption, which I know is wrong, that average rating is influenced by our player attributes. Is it possible, and how exactly, to do a statistic regression to estimate what the average rating of any player is going to be? I am not interested in the accuracy of the prediction, because the predictions is more likely going to fail, because there are other factors influencing the outcome. What interests me, and I don't know if that makes the question off topic sub forum material is the how to. For example is it going to be a linear regression? Is it going to be a logistic regression? is it going to be an ordinal regression? If it is a linear regression how do we explain negative coefficients if there is no negative effect on attributes? How much data sample size do we need for such a regression? Eventually, it is not the average rating I want to predict, but to find a way to estimate Y to create a "score" or "rating" for any given player, if I know their attributes, so that I can say player A is most likely better than player B. Has anyone attempted anything similar? Give details.