0
$\begingroup$

I've recently learnt the basics of regression as I progress through R and statistical modelling and approached a simple project of predicting NBA player salaries based off Age, Assists, Blocks, Turnovers, Points and eFG%.

Building a linear regression model just based off the stats (Assists, Blocks, Turnovers and Points) to project Salary is simple enough, but I'm not sure how to approach mixing in Age (which is from a range of ~20-35) and eFG% (which is a %) along with the stats. My first thought was feature scaling, but still pretty confused.

Thanks for any help. Much appreciated.

$\endgroup$

1 Answer 1

0
$\begingroup$

As far as I understand there is no problem with using your age variable in your regression.

If you really wanted to get a very specific answer you could create a factor with two groups under 25 and over 25, and then you could get a very specific regression coefficient for those two groups.

If you use the age as it is, I believe it will create a regression coefficient for each age, but I could be wrong as I don't have the data and I'm not doing it myself.

In essence, there is no problem with using data in the format of age in regression. In reality the dependent variable can be either quantitative or qualitative.

$\endgroup$
2
  • $\begingroup$ Using age as it comes will by default yield one regression coefficient for age in any regression software I have ever met. To get a separate coefficient for each age you would need explicitly to declare age to be categorical, a factor variable in one common terminology. $\endgroup$
    – Nick Cox
    Commented Nov 13, 2019 at 0:01
  • $\begingroup$ Thanks for the advice guys. How about adding the 'eFG' variable as it is a %? I feel since it's from 0 to 1 it seems to be tainting the model. Cheers. $\endgroup$
    – Kevin Chen
    Commented Nov 13, 2019 at 2:09

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.