I’m finding that most information on the internet pertaining to R squared seems to be arguments against it, and the issues with interpretting it as useful in prediction…So, can anybody explain what then it is godd for? I get that it quantified the amount of variation in Y that the regression line captures but what are the benefits/disadvantages of interpreting it in our forecasts?
Stop reading the internet. You’re not tested on what you find on the internet, nor its real-life application. You’re tested on the CFA curriculum.
That being said, the R-squared shows how much of the dependent variable is explained by the independent variable.
EG - A person’s weight is largely dependent on how tall they are. The taller they are, the more they will weight. Therefore, if you find the correlation between weight and height (which I calculated in another thread at .92), and you square the correlation (also known as R), then you get an R-squared of .85. That means that 85% of the variation in a person’s weight is explained by the variation in their height.
I understand what it means and how to calculate it,but that isn’t really what I was going for, thanks for the response. I’d like to know regardless
i use r-squared in everyday work so i’m talking about it in not-so-precise lanugage
in one factor regression, r-squared is usually very reliable to evaulate whether a regression equation is good or bad (capture enough variance). A low r-squared equation means the dependent variable has little correlation with indepedent variable (r-squared is exactly correlation squared in one factor regression)
in multiple factor regression, it is more tricky since adding one more factor will always increase r-squared. So sometimes people use adjust r-squared. Still, it is an useful figure to check wetheter your model is good or not.
one thing you need to know is r-squared is backward looking so theoractially it has nothing to do with focasting unless you believe forward data are from same distribution
Gotcha, and thanks- it made more sense once I got through multiple regression; I was confused as to how it could not be an indicator of a good model in a singe factor regression but I think I get it now.